COVID-19 and cycling: a review of the literature on changes in cycling levels and government policies from 2019 to 2022

ABSTRACT

This paper reviews 100 peer-reviewed articles and 12 non-refereed papers on COVID-19 and cycling published from March 2020 to January 2023. Overall, the studies suggest more increases than decreases in cycling, with some cities reporting large increases. However, there has been much variation among countries, cities, and specific corridors within cities as well as variation by gender, age, ethnicity, income group, trip purpose, and time period of the pandemic. The largest increases in cycling in 2020 were for recreation, exercise, and stress relief on weekends and weekday afternoons. By comparison, cycling to work, university, schools, and shopping generally declined. Most studies reported expansions or improvements in bikeway networks, often specifically related to COVID or accelerated due to COVID, and with a particular emphasis on low-stress facilities such as protected bike lanes, slow streets, car-free streets, and traffic calmed neighbourhood streets. Most of the studies examining the social equity impacts of COVID-related cycling policies found them to be broadly equitable across income, ethnic, age, and gender dimensions. Many studies recommended further expansion of low-stress, safer facilities in order to attract a broader cross-section of the population to cycling.

KEYWORDS:

• COVID-19
• bicycling
• cycling trends
• government policy
• active travel
• literature review

1. Introduction

In the three years since its onset, COVID-19 has claimed at least seven million lives and caused serious illness for hundreds of millions of persons (WHO, 2023). COVID-19 has also influenced many aspects of life throughout the world. Transport systems and travel behaviour have been affected by the public’s fear of COVID-19 infection and by government-mandated lockdowns, quarantines, travel bans, and closures. Those restrictions forced shifts to remote learning, working, shopping, and entertainment – most of which were temporary, but some of which have persisted to varying extents. That reduced travel overall, especially during COVID-19 peaks and before vaccines became available. The fear of contagion also influenced choice of travel mode, leading to declines in the usage of modes that require close proximity to other travellers in enclosed spaces, while favouring modes that enable distancing from other travellers – such as cycling (Buehler & Pucher, 2021a; de Vos, 2020; Habib & Anik, 2021; Möllers et al., 2022; Schaefer et al., 2021). Indeed, already in mid-2020, there were widespread media reports of a boom in bicycling and shortages of bicycles for purchase.

Assessing the effects of COVID-19 on cycling is important because cycling is a sustainable means of transport for travelling short to medium distances, especially in urban areas. Moreover, it has served as an alternative to public transport for some trips and for some people during the pandemic. In addition, increased cycling may have helped offset the reduction in physical activity due to the temporary closure of gyms, fitness centres, swimming pools, and athletic events during peaks of the pandemic.

Our review of the literature about government policies related to cycling examines the implementation of pro-cycling policies combined with restraints on motor vehicle use during the COVID pandemic, some of which would have been considered unlikely or even impossible in many cities under normal circumstances. The experience with a wide range of new or expanded measures during COVID revealed the effectiveness of low-stress, safe facilities such as protected bike lanes, off-road paths, slow streets, school streets, and traffic-calmed neighbourhoods to encourage a wider range of the population to ride bikes.

2. Studies included in this review

The purpose of this paper is to review the literature published on the possible impacts of the COVID-19 pandemic on cycling. The review examines 100 peer-reviewed journal articles and 12 non-refereed reports, papers, and datasets that were published between March 2020 and January 2023. Both kinds of studies are listed in Table 1, which is organised alphabetically according to the lead author’s last name. Studies that were not peer-reviewed are designated with asterisks following the authors’ names. We placed the table at the end of the full text to avoid interrupting the text with such a long table.

Table 1. Overview of 100 peer-reviewed and 12 non-refereed studies about bicycling and COVID-19 published between March 2020 and January 2023 (ordered alphabetically by lead-author last name). Reported for each publication are: study author(s) and publication year; type of study, data, and methods; time frame of study; number of cases; country/city; changes in cycling; government policies. Source: Summarised by the authors based on the articles listed in the table. The bibliographic information for each article is listed in the references at the end of the paper.

(1) Author(s) (publication year)	(2) Type of study, data, and methods	(3) Time frame of study	(4) Number of cases	(5) Country/ city	(6) Changes in cycling	(7) Government policies
Abdullah et al. (2022)	Factor analysis and discrete mode choice models.	May 2020 vs. pre-COVID	671 respondents	Major cities of Pakistan	The proportion of non-motorised modes increased during the pandemic.	Not measured, but recommendation to improve bikeway and pedestrian infrastructure.
Advani et al. (2021)	Alternative cycling levels estimated based on scenario simulations.	Unclear	Commuters in Delhi	Delhi, India	Simulated a six-fold increase in cycling if bicycle-friendly measures were implemented.	Not measured.
Aldred et al. (2021)	Spatial equity analysis on where LTNs were implemented during COVID-19.	March–September 2020	72 LTNs	London, UK	Not measured.	Low Traffic Neighborhoods (LTNs) in London have been broadly equitable. Residents of the most deprived neighbourhoods were 2.5 times more likely to live in a new LTN compared to residents of the least deprived neighbourhoods.
Ando et al. (2021)	Life Space Assessment modelling.	January 2021	11,886	Japanese cities (15)	During the COVID pandemic, bicycle users rode to more places, more often, and had greater mobility and accessibility.	Not measured.
Anke et al. (2021)	Online survey.	March–May 2020	4157	German states (16)	Cycling increased significantly in rural areas (22%–28%), but only slightly in the country overall (32%–33%).	Not measured.
Barbieri et al. (2021)	Negative binomial regression of data from online survey.	May 2020	9394	10 countries on 5 continents	Cycling declined in most countries	Not measured.
Batomen et al. (2023)	Descriptive analysis of survey responses from transit riders.	May 2020–April 2021	3500 regular transit riders	Toronto & Vancouver Canada	Transit riders cycled more during the pandemic.	Safer cycling facilities increase the likelihood for transit riders to cycle.
Becker et al. (2022)	Analysis of survey responses about pop-up bike lanes.	April 2019–March 2021	1661	Berlin, Germany	Cycling increased by an average of 73% in the 12 months since the pop-up bike lane was installed (April 2020–March 2021 compared to April 2019-March 2020).	Pedestrians (75%), cyclists (94%), and public transport users (79%) were supportive of pop-up bike lanes, while drivers opposed them (79%).
Bergantino et al. (2021)	Factor analysis and then ordered logit and probit analyses of online survey data.	2020	1163	Italian cities	Walking and cycling in 2020 post-lockdown remain 35% higher than pre-lockdown.	Survey found that 50% more bikesharing bicycles would induce 50% of the respondents to ride or ride more often. Recommends increasing bikesharing.
BTS (2022)*	Descriptive analysis of the operating status of bikesharing systems	2020	136	USA	Not measured.	From March to December 2020, 51 bikesharing systems ended operations permanently, 23 suspended operations temporarily, and 62 remained open.
Bucsky (2020)	Evaluations of trends using various data sources.	March 2020	Data from 5 bike counters and trip data from a bikesharing system.	Budapest, Hungary	Cycling and bikesharing usage increased as travel restrictions were imposed.	Not measured.
Budi et al. (2021)	Analysis of various factors that increased cycling during COVID-19.	March 2020–March 2021	34 studies	Indonesia	Health, environmental, media, and lifestyle factors helped increase cycling and bicycle sales.	Not measured.
Büchel et al. (2022)	Descriptive statistics and random forest regression; plus interviews with cycling experts.	2019 vs. 2020	33 bike counters	Basel and Zurich	During lockdowns there was more bike traffic in the afternoon and on weekends, but less cycling at other times. Post-lockdown cycling returned to pre-COVID levels in Basel and to slightly higher levels in Zurich.	Interviewed Swiss experts recommend implementing bike friendly measures to increase cycling.
Buehler and Pucher (2021a)	Review of existing studies and graphical portrayal of cycling trends over time and among countries.	2020 vs 2019	13 countries	11 EU countries, Canada and USA	Of 13 countries 11 reported net increases in cycling. All countries reported increases in weekend cycling, but 5 countries reported declines in weekday cycling. Recreational cycling increased, but cycling to work, school, and university decreased. Large variation by country, within countries, and by timing of COVID waves. Bicycle sales increased in all countries.	Provides a long list of measures implemented in various cities around the world based on national and international databases as well as information provided by individual cities. Many cities expanded cycling facilities, including both permanent and experimental pop-up bike lanes. Other measures included expanded traffic calming, slow streets, and car-free zones.
Buehler and Pucher (2022)	Updated literature review and updated graphical portrayal of cycling trends over time and among 13 countries, supplemented by detailed case studies of 14 cities.	2020 and 2021	13 countries and 14 cities	13 countries and 14 cities in Europe and North America	Great variability among countries and over time in month-by-month changes in cycling levels. Large declines in cycling during lockdowns, but large increases after lockdowns, with the timing varying by country corresponding to COVID waves. Cycling increased in all 13 countries on weekends; but declined in 4 countries on weekdays. Of the 14 case study cities, cycling increased in 6 cities, decreased in 3 cities, and remained roughly the same in 5 cities.	COVID-19 led to expanded and improved cycling facilities financed by increased government funding. An important emphasis in that expansion was the provision of low-stress, protected bike lanes. Other measures included expanded traffic calming, slow streets, reduced city wide speed limits, and car-free zones.
Cao and Wang (2022)	Development of a bikesharing demand model.	January 2020–April 2021	38.33 million bikesharing trips	Shenzhen, China	COVID lead to a surge in bikesharing – especially around rail stations.	Cutting public transport service pushed riders to bicycles.
Campisi et al. (2020)	Ordinal regressions on stated preferences from a survey.	March–April 2020	431 individuals	Sicily, Italy	61% of respondents predicted more cycling after the pandemic.	Recommendation: Bicycling promotion must be part of future mobility measures implemented in Italian cities.
Carrese et al. (2021)	Transport system modelling to predict future travel demand.	March–June 2020	Unclear	Rome, Italy	In the historic centre of Rome, cycling declined by 89% during lockdown. Even after the lockdown, cycling was 68% less in early May than pre-COVID.	Not measured.
Chen et al. (2022)	Analysis of bikesharing trips.	January–December 2020	Capital bikesharing data	Washington, DC	Large decline in bikesharing during COVID lockdown (March/April), then increasing slightly during the rest of 2020, but remaining lower than in 2019.	Not measured.
Combs and Pardo (2021)	Catalog of COVID-19 measures that reallocated roadway space to promote walking and cycling.	March–August 2020	Over 1000 COVID measures from around the world	Cities in 60 countries on 6 continents	Not measured.	Over 1000 roadway space reallocation measures implemented: shift of motor vehicle traffic lanes to cycling or walking (13%); full or partial street closures (22%); and expanded bikesharing (8%).
Costa et al. (2022)	Multinomial logit model on stated cycling levels based on a 1-min questionnaire in 5 locations	May 2020	493 individuals	Lisbon, Portugal	Compared to pre-pandemic cycling, 40% cycled more and 40% maintained their cycling frequency. After lockdown, trips for recreation and exercise increased while trips for work and school decreased. Those with little pre-pandemic cycling increased their cycling the most.	Recommendation: Authorities should build more bikeways and provide other incentives for cycling.
Cusack (2021)	Logistic regression of online survey data.	June–August 2020	213 including 65 cyclists	Philadelphia	Lower likelihood to bike to work for non-whites, those with time constraints, and concerns about safety; more cycling for those most concerned about COVID.	Safe cycling infrastructure would increase cycling.
de Haas et al. (2020)	Descriptive analysis of national mobility panel to identify COVID impacts.	2020 vs 2019	2500	Netherlands	The mode share of cycling remained constant, but average bike trip distance increased by 30%, mainly due to longer round-trips for recreation and exercise. The survey also found that 20% of respondents expected to cycle more after the pandemic, compared to only 3% anticipating less cycling.	Not measured.
de Séjournet et al. (2022)	Regression analysis based on counter and Google Community Platform data.	January 2020–December 2021	Data from bike counters	Brussels, Belgium	Amplifying a pre-pandemic trend, cycling increased significantly on weekdays (+63%) and even more on weekends (+189%).	Closures of workplaces and grocery stores were not correlated significantly with cycling levels, but school closures were related to less cycling.
de Vos (2020)	Editorial.	April 2020	None	No specific location	Walking and cycling might increase, because they facilitate social distancing. People might walk and cycle more recreationally, because of the reduction of out-of-home activities during COVID.	Not measured.
Doubleday et al. (2021)	Linear regressions of bike count data.	May, June and August 2020	Counters (4 in Seattle; 5 in NYC, 3 in Houston)	Houston, New York, Seattle	Much variation among the three cities, as well as within each of the three cities by counter location. Overall, cycling increased in Houston and fell in New York, probably due to its strict lockdown over this period.	Not measured.
Duren et al. (2021)	Multi-level regression of online survey data with post stratification.	June 2020	2011	US wide	Not measured.	Limited public support for car restrictions or roadway reallocations to promote safer bicycling and walking.
ECF (2022)*	Database of kilometers of bike-friendly measures announced and implemented as well as funding allocated for bicycle promotion in European countries.	2020–2022	44 countries and 94 large cities	European countries and cities	Not measured.	Since March 2020, roughly 2600 km of bikeway infrastructure, traffic calming, and car-free areas were announced by local, regional, state, and national level governments. Of those, about 1500 km were implemented by September 2022. During the same time period 1.7bn Euros have been allocated for bicycling in Europe–in the form of additional funding for infrastructure, tax incentives/reductions for cycling, subsidies for bicycle purchases and repair, and promotional campaigns.
ECF & Eco Counter (2022)*	Analysis of bike trip data from 170 Eco-Counters along European cycling routes.	2019–2022	170 counters	Long-distance cycling routes throughout Europe.	Over the period 2019–2022, cycling increased on weekends as well as on weekdays. Over the previous periods 2019–2020 and 2019–2021, cycling increased on weekends, but declined during weekdays.	Not measured.
ECG (2020)*	Trend analysis of automatic counter data.	2019–2020	15 automatic counters	East Coast Greenway trails in Maine, Connecticut, Pennsylvania, and Virginia.	From 2019 to 2020, walking and cycling increased by more than 50% at various counter sites along the ECG, but with considerable variation among sites.	Recommendations for expanded, improved, and better connected greenway facilities – based on large increase in usage during COVID.
Echaniz et al. (2021)	Multinomial logit analysis of stated preferences on travel before, during, and after lockdowns.	April 2020	336 full & 478 partial responses	Spain	Reductions in cycling during lockdown, but increases in cycling after lockdowns.	Not measured.
Eco-Counter (2023)*	Descriptive analysis of representative samples of bike counters for each country.	2022, 2021, and 2020 vs 2019	575 counters	11 EU countries and 8 regions of the USA and Canada	From 2019 to 2020, cycling increased by 8% in Europe (+3% weekdays; + 23% on weekends) and by 16% in the United States (+29% on weekends; + 10% on weekdays). Cycling on weekdays decreased in the morning commute hours, but increased in the mid to late afternoon.	Not measured.
Ehsani et al. (2021)	Revealed preference survey of cycling levels before and during pandemic.	June 2020	2011 adults (see other study with this survey)	US nationwide	No significant change in reported bicycling overall, but share of bike trips increased because travel declined sharply. 25% of respondents indicated that they anticipated cycling more after COVID (33% in suburban areas vs. 11% in urban areas).	Not measured. Recommendation that investment in safe bicycling infrastructure could facilitate the anticipated increase in bicycling.
Eisenmann et al. (2021)	Binary logistic regression analysis of travel survey data.	March and April 2020	804 respondents 18 or older	Germany	During lockdown, the share of individuals who cycled for most of their trips increased from 6% to 9%. One in five respondents rode a bicycle both before and during the lockdown.	Not measured.
Evenson et al. (2023)	Information from websites and interviews.	July 2020–December 2021	Websites of 314 communities and 12 interviews	US cities with 100,000 + population	Not measured.	Municipalities larger than 165,000 implemented twice as many measures compared to smaller jurisdictions. One in 6 measures aimed to reduce active travel (e.g. access limits to reduce crowding).
Firth et al. (2021)	GIS and census data based analysis of socio-spatial differences in the distribution of street reallocations.	April–December 2020	2 cities	Seattle, Washington, USA and Vancouver, British Columbia, Canada	Not measured.	Vancouver (40 km) and Seattle (20 miles) reallocated streetspace for cyclists and pedestrians in residential low traffic streets outside of the city centre. More streetspace reallocations in areas where people of colour, particularly Black and Indigenous people, lived.
Fischer and Gopal (2021)	Descriptive analysis of online-survey data.	May–June 2020	428 respondents	136 urban areas in 39 countries	For 14% (20% for men) cycling was the main sources of physical activity. 46% changed their cycling route to use green streets.	One-third of respondents in low-income countries did not have access to greenspace compared to 8% in high income countries.
Fischer and Winters (2021)	Analysis of street-reallocations during COVID.	April–August 2020	3 cities	Canadian cities	Not measured.	All cities reallocated streetscape for walking and cycling: Halifax (17.2 km), Victoria (6.4 km), and Kelowna (0.7 km). The bicycle-related interventions in Halifax and Kelowna were concentrated in higher-income, non-minority areas, while those in Victoria were focused on lower-income, minority areas.
Fischer et al. (2022)	Analysis of changing spatial patterns of bicycling using Strava data and GIS.	April to October 2019 vs 2020	Strava data	Vancouver	Across all subsets of the population, the largest increases in cycling were on high-comfort facilities, such as protected bike lanes, off-street paths, and slow streets with restricted car traffic.	Recommendation for low-stress cycling facilities such as protected bike lanes and off-street trails.
Fuller et al. (2021)	Survey of Australians who changed their cycling behaviour during the pandemic lockdowns.	August–September 2020	444 cyclists (but 17% restarted cycling)	Australia	85% increased or maintained their cycling levels during lockdown vs 15% who decreased cycling. Most of the increase was for recreation, exercise, socialising, and wellbeing, but not for transport. Reduced motorised traffic volumes were more important for cycling than pop-up bike lanes.	Recommendation for more low-stress cycling facilities.
Gladwin and Duncan (2022)	Descriptive analysis of a survey of travel behaviour and attitudes of seniors (65+).	February–April 2021	198	Tallahassee, Florida	Cycling frequency decreased but average trip distance increased. Cycling during the pandemic was associated with enjoyment, self-esteem, and increased freedom. Of those who increased their cycling during the pandemic, 80% plan to continue cycling after the pandemic.	Recommends government programmes to promote cycling among older adults.
Glaser and Krizek (2021)	Inventory of COVID measures for 55 largest US cities; trying to categorise the measures; 30 cities qualified as “transition experiments”.	June and December 2020	55 cities	55 U.S. cities	Not measured.	The most common COVID measures in June 2020 were “open streets” or “slow streets” (30 of 55 cities). A 6-month follow-up in December showed that 50 percent of these measures (15 cities) were still in progress, including 6 cities that expanded or made permanent some experiments.
Habib and Anik (2021)	Analysis of tweets using text mining and topic modelling techniques.	May and June 2020	15,776 tweets	English language tweets	Main shift from public transport to private cars, bicycling, or walking. Bicycle-related tweets discussed shifts from other modes, public health benefits, and improved bicycling infrastructure.	Cycling identified as a useful travel alternative during COVID-19 as well as after the pandemic to reduce traffic congestion.
Harris and McCue (2023)	10 semi structured interviews with experts and government officials.	February 2020–March 2021	10 in-depth interviews	Sydney, Australia	Between July 2020 and April 2021, more than 500,000 trips were taken on six pop-up cycleways. One of those became the most heavily used cycleway in the city, averaging a 500% increase in cycle trips.	COVID-19 pandemic created a policy window for rapid implementation of bicycle–friendly policies. Six pop-up cycleways were constructed within 3 months that constituted 38% of the city’s protected cycleway network.
Hassen (2022)	Theoretical exploration of health inequities during COVID-19.	April 2022	None	Toronto, Canada	Not measured.	Protected bikeway networks in under-resourced communities helped improve health.
Heydari et al. (2021)	Bayesian second-order random walk time-series model to compare observed number of cycle hires and hire time with their respective counterfactuals without the pandemic.	March–December 2020	Data on bikesharing docking stations	London, UK	Cycle hires declined during lockdown in March and April 2020, but rebounded in May 2020 to prepandemic levels after lockdown was lifted. After the lockdown, bikes were hired for longer duration, partly due to a shift from relatively longer public transport trips.	Not measured.
Hong et al. (2022)	Crowdsourced cycling data and regression models to examine the extent to which cycling intensity for non-commuting purposes changed with different types of cycling infrastructure.	July 2019– March 2020	Strava data	Glasgow, Scotland	Large increases in recreational and exercise cycling along rivers, parks, and other off-road facilities. Cycling also increased on roadways with low traffic volume. Decrease in cycling to work, university, and school due to lockdowns and remote working and learning.	Recommendation for expanded network of off-road cycling facilities and low-stress, protected bike lanes on roadways.
Huang et al. (2023)	Analysis of GPS tracking data for weekday trips of 252 study participants.	September 2019– November 2020	GPS tracking data for 252 participating individuals making 162,692 trips	Switzerland	Compared to pre-COVID, cycling mode share of all trips roughly doubled both for persons working from home and for persons still making the trip to work: 4%–8% vs 3%–7%.	Study implicitly encourages policies to facilitate working from home because it leads to more active travel for non-work trips while reducing negative impacts through reduction of work commute trips, most of which are by private car.
Infrastructure Australia (2020)*	A review of COVID impacts on a wide range of topics, based on information from wide variety of sources.	Mid 2019–mid 2020	3 cities	Australia, with cycling data from Sydney, Brisbane, and Melbourne	Increased cycling levels in Melbourne, Sydney, and Brisbane from mid-2019 to mid-2020.	Brisbane, Melbourne, and Sydney all reported expanded cycling facilities, especially pop-up bike lanes.
Irawan et al. (2022)	Survey of cyclist attitudes toward future cycling due to COVID experiences.	December 2020 to February 2021	473 cyclists	Yogyakarta, Indonesia	Cyclists said they were more likely to cycle in the future due to COVID.	Actual government policies during COVID not measured, but authors recommend promoting cycling based on its environmental and safety benefits.
Jobe and Griffin (2021)	Survey of employed vs. unemployed residents about whether they increased bikesharing during COVID-19.	June–July 2020	101 former or current bikesharing users	San Antonio, TX	Bikesharing use increased among the unemployed but decreased among the employed. Largest increase in bikesharing use among infrequent users (1–2 rides per month).	Not measured.
Kazemzadeh and Koglin (2021)	Analysis of qualitative interviews with e-bike non-users.	January and April 2020	23 e-bike non-users	Lund, Sweden	Respondents stated an increased attractiveness of e-bikes as possible substitutes for public transport after the pandemic.	Not measured.
Kearns and Wright (2022)*	Analysis of automatic counter data at various locations throughout North Carolina, including roadways and recreational trails.	March–September, 2019 vs. 2020	over 70 automatic counter locations	Over 70 automatic counter sites throughout North Carolina	Large increases in cycling 2019–2020 on off-road recreational greenways but declines in cycling in and to commercial areas and university campuses (Geiger et al., 2021). The largest increases in cycling were on weekends and on weekday afternoons and early evenings.	Not measured.
Kraus and Koch (2021)	GIS-based regression analysis.	2020 vs 2019	106 European cities, 736 automatic bike counters.	106 European cities	Installation of pop-up bike lanes led to increases in cycling levels ranging from 11% to 48% during COVID.	European cities installed an average of 11.4 km of pop-up bike lanes per city.
Ku et al. (2021)	Change in bikesharing use during COVID.	January–June 2019 vs. 2020	Bikesharing data	Seoul, Korea	Bikesharing usage increased during COVID.	Not measured.
Kubaľák et al. (2021)	Descriptive analysis of bikesharing trip data.	May–December 2020 vs 2019	268,978 bikesharing trips	Košice, Slovakia	Bikesharing rentals decreased by 46% in 2020 vs. 2019 due to remote work and school as well as fear of infection. Average time per bike rental increased.	Not measured.
Kurkcu et al. (2021)	Analysis of cycling levels measured by automated counters.	2018 vs 2020	Four bike counters	Colorado	Results varied greatly by location.	Not measured.
Kutela et al. (2022)	Text mining and network analysis of 483 measures, with logistic regression to estimate the likelihood of COVID transport measures to become permanent.	March 2020–January 2022	483 measures	550 cities in 60 countries	Not measured.	75% of 483 measures were short-term, while 12% were permanent, and another 12% were not yet determined. In both the network analysis and logistic regression, bike lanes were the most likely to be made permanent, followed by full-street closures (open streets).
Landgrave-Serrano and Stoker (2023)	Quasi-experimental research design to measure walking and cycling on Slow Streets.	August 2020 and March 2021	15 Slow Streets and 15 control streets	Tucson, Arizona	Slow Streets more than doubled cycling and walking compared to control streets–in spite of the hot climate and lack of shade in Tucson.	Local governments should consider implementing Slow Streets to promote walking and cycling.
Li and Zhao (2022)	Fixed-effect logit models to estimate changes in cyclist crashes before and during the pandemic	2013–June 2021	Over 4 million datapoints from CitiBike	New York City	The number of overall crashes declined, but crash severity increased. The number of cyclists killed or injured per crash tripled compared to years before the pandemic.	Local governments should identify hot spots for cyclist crashes (as in this study) and focus coordinated safety measures at those locations, including stricter speed enforcement.
Li et al. (2021)	Changes in bikesharing use.	January–June 2020	Data from 761 docking stations	Greater London	Significant spatial variation in changes in bikesharing use from one part of London to another. Increases were near parks or recreational areas, but decreases along commutation corridors. Large declines during lockdowns followed by large increases afterwards.	Not measured.
Lin et al. (2021)	ARC GIS analysis of accessibility.	2020 vs. pre-COVID	59,554 links	Toronto, Canada	Not measured.	Expansion of low-stress cycling facilities during COVID greatly increased accessibility for much of Toronto’s population.
Lindsey et al. (2022)	Changes in cycling levels over time at various bike counter locations.	2017–2020	57 automatic bike counters	Minnesota	20% increase in cycling on off-road trails and paths, 2019–2020; 14% decrease in cycling on roadways over same period. Significant variation among locations and facilities of each type.	Not measured, but with recommendation that governments invest more in automatic counters to track trends over time.
Lopetrone and Biondi (2022)	Survey analysis of driver behaviour.	March 2020–January 2021	103 survey respondents	Windsor, Canada	Overall perception of more aggressive driving, speeding, distracted driving during COVID, thus endangering cyclists.	Not measured.
M-Five GmbH (2021)*	Cross-tabulation analysis of Eco-Counter data from Munich, Cologne and Berlin, comparing the period January-June for 2019, 2020, and 2021.	2019–2021	3 cities	Munich, Cologne and Berlin, Germany	Over the period January to June, total cycle trips increased in all three cities from 2019 to 2020: Berlin +19%, Cologne +11%, Munich +20%. Over the longer period 2019–2021, cycling increased by 9% in Berlin and 20% in Munich, but declined by 4% in Cologne.	Not measured.
Möllers et al. (2022)	Portrayal of cycling trends and regression analysis of various factors affecting cycling levels based on automatic counter data.	January 2019–June 2020.	10 cities	10 German cities	In 8 of 10 German cities, cycling increased during COVID, mainly due to shifts from public transport to cycling. Only 2 university cities (both with very high pre-COVID bicycle mode shares) reported decreases in cycling due to remote learning, both at universities and schools. Total trips declined sharply in all cities; thus, the modal share of cycling increased in 9 of 10 cities. The largest increases were in cities with relatively low levels of cycling pre-COVID and in cities with stricter lockdowns.	Not measured.
Molloy et al. (2021)	App based (GPS) tracking and online surveys.	March–August 2020 vs 2019	1439	Switzerland	Roughly a doubling in cycling, and more than a doubling in cycling mode share due to decline in total travel. Most of the increase was for recreation or exercise, with decreases in work commutation. Largest increases on weekends, holidays, and weekday afternoons.	Not measured, but recommendations made for Swiss cities to provide more cycling infrastructure.
Monfort et al. (2021)	Analysis of data from 16 bike counters and supplementary cyclist injury data.	2013–2019 vs 2020	16 bike counter sites, 9.15 million total bike trips counted	Arlington, VA	In 2020 midday bicycle traffic increased by +76% compared to 2019, but morning bicycle traffic fell by – 49%. Evening bicycle traffic increased on off-road, multi-use trails (+6%), but decreased for on-road lanes (−27%). In 2020, the bicycle injury crash rate per cyclist dropped by 28% – partly due to more cycling on off-road trails.	Recommendation that governments provide safer, protected on-road cycling facilities and more off-road trails because they are safer.
Monterde-i-Bort et al. (2022)	Analysis of internet survey data gathered via social media “snowball” method.	Spring vs autumn 2020	636	10 countries	No significant changes in cycling levels between 2019 and 2020, also not between spring and autumn 2020.	Not measured.
Morzynski et al.* (2020)	Analysis of app-based data from Streetlight.	May–July 2020	100 MSAs	100 largest US MSAs	Between July 2019 and July 2020, cycling increased in over 70 MSAs, especially in smaller MSAs, where cycling rose by more than 50%. Overall, however, cycling increased by only 12% on average for all 100 MSAs, due to declines in large MSAs.	Not measured.
Musulin et al.* (2021)	Analysis of COVID cycling trends from January 2019 to July 2021 using app-based tracking data from Streetlight.	January 2019– July 2021	100 metro areas	100 metro areas in the USA	Total cycle trips in the US in July 2021 were roughly the same as in July 2020 (height of the pandemic), both of which were 10% higher than in July 2019 (before the pandemic). Decreased cycling in West Coast and Midwest metros in 2021 versus 2020, but increased cycling in Southwest metros for the same time period. In July 2021, 81 of 100 metro areas, had higher cycling levels in July 2021 than in July 2019. The largest increases were in cities with relatively low levels of cycling pre-COVID.	Not measured.
Nguyen et al. (2021)	Logistic regression on data from a survey based on parent-based convenience sample.	May 2020	472 school-aged children	Hanoi, Vietnam	Bicycling and e-bikes (2%–11%) increased for the school commute after the lockdown (mainly replacing walk trips).	Recommendation: Policies, tools, and programmes that limit car and motorcycle ownership and use are needed.
Nguyen and Pojani (2022)	Binary and ordered logit models on likelihood to (1) start cycling during the pandemic, (2) trip purpose, and (3) cycling frequency based on survey data.	March–April 2021	356 recreational cyclists	Hanoi, Vietnam	25% started cycling recreationally during the lockdown. 75% noticed increase in cycling. Men, those with children, and those not working or studying most likely started cycling. Preferred time for recreational cycling was early in the morning.	Government policies had restricted walking in popular parks during lockdowns. Recommendation: Multi-use paths in parks should be created and car access should be restricted in parks to protect cyclists.
Nikitas et al. (2021)	Wide-ranging review to identify papers with case studies of cities that promoted cycling during COVID.	2019 vs 2020	about 50 cities around the world	39 cities on 5 continents	In general, large increases in cycling in almost all cities examined.	Recommends pro-cycling measures that should be more widely adopted, and to a greater extent in each city, in the future, to build on the success of such measures during COVID.
Nikiforiadis et al. (2020)	Descriptive, principal component, and ordinal regression analysis of stated preferences about bikesharing.	June–July 2020	223 individuals	Thessaloniki, Greece	COVID-19 will not affect bikesharing use in the future. Bikesharing was seen as safer than public transport.	Using mobile apps for bikesharing rental and disinfecting bicycles could have attracted more riders during COVID-19.
Nordengen et al. (2021)	Calculation of a national cycling index based on counter data.	2018–August 2020	89 bicycle counters	Norway	Cycling trips increased by 11% between 2018 and 2020 in Norway with a 23% increase in southern Norway, but a 8% decrease in northern Norway.	Not measured.
Ohlund et al. (2022)	Examines the design, methods, and community response to a new bike lane.	May–July 2020	1 new bike lane	Zapopan, Mexico	Cyclist counts increased by 26% along a corridor with a new bike lane between May and August 2020 (by 32% compared to 2016 levels).	The COVID-19 pandemic acted as a catalyst for quickly constructing a lane that had long been planned.
Ozbilen and Akar (2023)	Structural-equations modelling to link neighbourhood characteristics, well-being, and perceived risk of COVID infection for active travellers and shared micromobility users.	April 2020–February 2021	1972 survey responses	Columbus, Ohio	Residents in compact, mixed-use neighbourhoods perceived walking and cycling as less risky for COVID infection and reported greater subjective well-being.	Not measured.
Paiva et al. (2022)	ANOVA to compare trip origins and destinations before and after COVID restrictions using app-based Google Community Reports data.	May–June 2021 vs. January–February 2020	Large Google Community Reports dataset.	Spain and Portugal	Significant increase in trips within parks and decrease in trips to workplaces.	Recommends building more cycling infrastructure to increase cycling.
People for Bikes (2021)*	Descriptive analysis of national survey.	2020	2803	USA	4% of respondents said they rode a bicycle for the first time ever in March 2020 or for the first time since over a year ago. New cyclists reported five main reasons for riding a bike (not mutually exclusive): stress relief and mental health (58%), exercise and physical fitness (57%), socialising with friends and family (43%), relaxation (37%), and getting outdoors (33%).	Approximately 200 U.S. cities reallocated streetscape to accommodate increased outdoor activity, including cycling.
People for Bikes (2022)*	Measured expansion of cycling facilities.	2020–2022	5 cities	5 US cities	Not measured.	From 2020 to 2022, Austin (TX) built 115 miles of bicycling facilities and is planning to build another 250 miles by 2025. Between 2020 and 2022, Pittsburgh (PA) built 65 miles, Providence (RI) 65 miles, New Orleans (LO) 27 miles, and Denver (CO) 150 miles.
Porter et al. (2021)	Analysis of mobility diaries about everyday transport challenges by young women in poor peripheral communities.	Early 2019–March 2021	419 entries from 22 mobility diaries	3 African cities (Abuja, Cape Town and Tunis)	Not measured.	There is potential for prioritisation of Non-Motorised Transport interventions that benefit women through promoting women’s cycling.
Qu et al. (2022)	Multinomial logistic regression analysis of survey data	November 2020	1000	Michigan, USA	Overall decline in already very low levels of cycling in Michigan, but larger decreases for lower-income, lower-education, minority, female, and older persons.	Recommends government policies to encourage more walking and cycling among non-white, less-educated, older, and female persons, including better cycling and walking infrastructure and slow streets to improve safety.
Rérat et al. (2022)	Logistic regression analysis of data from stated preference survey of cyclists.	June–August 2021	1401	Geneva and Lausanne, Switzerland	44% of Lausanne cyclists rode more after the pandemic than in 2019, while 42% cycled the same amount. 27% of Geneva cyclists rode more after the pandemic than in 2019, while 58% cycled the same amount. Declines in work commute trips in both cities were more than offset by large increases in recreational and exercise cycling. Cycling lanes built during COVID were perceived to be lower-stress, safer, more convenient, and faster, thus resulting in a changed pattern of cycling routes.	The pandemic accelerated existing cycling infrastructure plans; both Geneva and Lausanne each built 7.5 km of additional cycle lanes in spring 2020. New cycle lanes were created from car traffic lanes in Geneva, but from car parking spaces in Lausanne.
RTC (2021)*	Descriptive analysis of automatic counter data for off-road rail trails.	2020	43 counters	USA	From 2019 to 2020 more cycling and walking at all hours of the day and on all days of the week. Largest increase in spring 2020 during the first weeks of the pandemic; smaller increases later in 2020.	Not measured.
RTC (2023)*	Regularly updated database of extent off-road trails	December 2019–September 2022	Hundreds of greenways	USA	Not measured.	15% increases in kilometers of off-road greenways in USA
Saatchian et al. (2021)	Descriptive analysis and structural equations modelling of stated preferences about bikesharing	2020	310 male bikesharing members	Mashhad, Iran	Bikesharing is effective in improving perceived health and well-being during COVID-19.	Not measured.
Sangveraphunsiri et al. (2022)	Non-negative Tucker decomposition to analyse bikesharing trips before and during the COVID-19 pandemic.	2018–2020	Bikesharing trips between 50 stations on 1010 days	Bangkok, Thailand	Work-related bikesharing trips fell significantly during the pandemic, but average trip duration increased from 9 to 14 min. Bikesharing recovered more quickly from COVID-19 trip reductions than other modes of transport.	Not measured.
Schaefer et al. (2021)	Multiple regression analysis of survey data to identify factors causing shifts away from public transport to other modes, including cycling.	June 2020	Over 4000 survey respondents	Metro region of Hannover, Germany	Mode shift to cycling from public transport for short trips using the Stadtbahn (tram), but not for longer trips on suburban trains, for which cycling was identified as an important feeder mode.	Recommendation for low-stress, protected cycling facilities in general, and improved cycleways leading to suburban train stations.
Schweizer et al. (2021)	Four different generalised linear models used to analyse Strava cycling data in public green spaces.	March–June 2020	Parks in 15 large cities and 7 rural areas	Parks in large cities and rural areas in Germany	55% average increase in recreational and exercise cycling in parks of 15 large cities from March to June 2020, but no significant changes in parks in rural areas.	Recommendation for improved off-road cycling facilities to promote cycling for physical fitness, health, and stress relief.
Scorrano and Danielis (2021)	Analysis of survey data to determine change in mode share of trips from 2019 to 2020.	April–August 2020	315	Trieste, Italy	Small increases in cycling and bikesharing between 2019 and 2020.	Recommends better cycling infrastructure to encourage more cycling.
Shaer and Haghshenas (2021b)	Logistic regression of survey data from older adults.	2020	315 surveyed participants	Isfahan, Iran	Doubling in cycling mode share of trips among 60 + seniors, from 9% to 18% of all trips. Older adults also reported a 50% increase in the duration of bike trips (10–15 min). Almost all of the increased cycling was by men and not women, who increased their walking instead.	Recommendation for governments to improve quality and safety of walking and cycling routes.
Shibayama et al. (2021)	Binary logistic regression analysis of online questionnaire of work commuters.	March–May 2020	11,000 survey responses	14 countries	Large reductions in work commuting by bicycle in 9 countries during the height of the COVID pandemic, with only slight increases in 3 countries.	Not measured.
Shirgaokar et al. (2021)	Analysis of crowdsourced dataset from Twitter of perceptions/opinions about measures during COVID.	April–July, 2020	292,108 tweets	USA and Canada	Not measured.	Recommendation to follow-up on the reallocation of street space during COVID, especially for uses other than motor vehicle traffic and parking.
Song et al. (2022)	Flow dynamic analysis to evaluate changes in number and direction of bikesharing trips.	December 2019 –July 2020	30-second interval records for bikesharing trips for 8 months	Singapore	Bikesharing ridership increased 150% during the lockdown. Cycling increased in the city centre and at waterfront areas.	Bikesharing is recommended as an alternative when public transport systems are closed, such as during the COVID-19 pandemic.
Strömberg and Wallgren (2022)	Analysis of interviews with people who started cycling during COVID, identifying barriers to cycling.	Mid to late 2020	12 interviews	Gothenberg, Sweden	Not measured.	Recommendations to overcome barriers to cycling by building more protected bike lanes, lockers and indoor storage at workplaces, as well as informational campaigns for current and future cyclists.
Sung (2023)	Bayesian structural time series models and causal impact inference to estimate changes to ridership during COVID-19.	January 2017–December 2021	1826	Seoul, South Korea	Overall bikesharing ridership was unaffected by the pandemic, however trip purposes shifted from transport to exercise and leisure.	Recommendation: Infrastructure investment is required to make public bicycle travel more resilient against the pandemic.
Sunio and Mateo-Babiano (2022)	Online ethnography based on information from Facebook, other online groups, media reports, and webinars, supplemented by academic and professional policy papers.	March-August 2020	3 metro areas	3 metro areas in the Philippines	Authors report a “cycling boom” in Metro Manila but do not provide any quantitative evidence.	National government allocated $23M for a 500-km bike lane network that was completed in 2 years in the 3 major metropolitan areas, including Metro Manila. Cycling activist groups played a key role in forcing the rapid implementation of the government's plan. Academic and professional experts also recommended implementation.
Tarasi et al. (2021)	Descriptive analysis and various regression models based on survey data.	January, February vs March, April, June 2020	308 and 193 participants in two samples	Rethymno and Chania, Crete	Share of respondents who did not ride bicycles increased 9% during the lockdown, fell after that, but increased again by 11% between April and June.	The lack of infrastructure and cultural barriers limit cycling.
Teixeira et al. (2021, 2022)	Travel behaviour survey of bikesharing users.	2020	215	Lisbon, Portugal	Large fall in bikesharing during lockdown due to closed workplaces, schools, universities, and shops. Large increases in bikesharing use after lockdown, mainly from former public transport riders.	Not measured.
Thomas et al. (2022)	Multilevel regression analysis to estimate the equitable distribution of school street closures.	March 2020–June 2022	420 school street programmes	London, UK	Not measured.	School Streets have been equally distributed across several socio-demographic indicators, they are less likely to benefit schools in car-dominated areas with poor air quality, and their spatial distribution is highly unequal.
Thombre and Agarwal (2022)	Descriptive analysis web-based survey before and during COVID and simulation model for one city.	March–July 2020	470	India and City of Patna	Increase in cycling in the trip distance range of 5–10 km in India. For the city of Patna, India a simulation model predicted that building a bicycle superhighway would increase bicycle mode share for work trips from 31% to 44%.	The authors recommend greatly increased expansion and improvement of bicycling infrastructure in Indian cities, especially express bikeways separated from motor vehicle traffic.
Tiako and Stokes (2021)	Review of bikesharing trends, programmes implemented, and literature on bikesharing equity.	2020 - 2021	Bikesharing systems in the US	U.S. cities with bikesharing systems	Not measured.	Many bikesharing companies offered free or low cost memberships to essential workers (who were more likely Black or LatinX). However, inclusion criteria of these programmes for essential workers were arbitrary and there were comparatively fewer docking stations in minority neighbourhoods.
Valenzuela-Levi et al. (2021)	Scenario analysis to study potential bike access to jobs (<30 min or 7 km) based on secondary travel survey, census, and housing data.	Data from 2012 and 2017	Large, existing government datasets on travel, census, and housing	Santiago de Chile	Among those who kept commuting in a district near to Santiago’s CBD (Providencia) cycle-commuting increased from 25 to 32% during the pandemic.	Locating subsidised housing in cycling distance to job concentrations can help increase equity in job access during a pandemic when lower income workers still need to go to work. Recommendation: The authors advocate a post-COVID situation with affordable housing located near job concentrations so that trip distances are short enough to be covered by bike.
Vallejo-Borda et al. (2022)	Structural Equation – Multiple Indicator Multiple Cause (SEM-MIMIC) model to explain shifts from public transport to active travel or driving.	September– November 2020	3721 survey responses	Bogota, Buenos Aires, Lima, Quito, and Santiago	During COVID there was a lower probability of shifting from public transport to active travel compared to driving a car–in particular for women and seniors.	Not measured.
Waitt et al. (2023)	Ethnography combined online semi-structured interviews with video methods and follow-up online conversations about riding e-bikes during lockdown.	July-December 2020	30 participants	Sydney, Australia	COVID-19 disrupted commuting routines and allowed respondents to ride e-bikes, due to increased safety concerns about public transport and decreased safety concerns on roads due to decreased car traffic.	Cycling friendly policies mattered less than the disruption caused by COVID.
Waitt and Stanes (2022)	Semi-structured interviews, sketches, and follow-up conversations around self-generated video-recordings of cycle rides.	July-December 2020	3 participants	Sydney, Australia	COVID-19 disrupted commuting routines and allowed respondents to return to riding a bicycle for their commutes, due to increased safety concerns about public transport.	Recommendation: Future policy must address gendered mobility hierarches and increase safety and cyclists’ right to the road.
Wang et al. (2023)	Univariate and multivariate linear models with Walk Score, Transit Score, and Bike Score as independent variables and COVID-19 infection rates as dependent variables.	April 2020–April 2021	72 cities	72 cities in Massachusetts, USA	Higher Walk Score, Transit Score and Bike Score were shown to be protective against COVID-19 transmission.	Authors identify the built environment as an additional layer of social determinants of health.
Xin et al. (2022)	Spatiotemporal GIS analysis using bikesharing data.	January–April 2019 vs January–April 2020	9,143,730 trips	New York City boroughs of Manhattan, Brooklyn, and Queens	Relative to March and April 2019, sharp decline in CitiBike usage in March and April 2020, during the worst of COVID crisis in NYC. Especially steep decline in morning peak hour trips and in Lower Manhattan due to reduction in work commuting. Largest reduction in short distance rides. Less reduction in longer distance rides.	Not measured.
Zafri et al. (2021)	Descriptive and multinomial regression analysis to identify mode usage before and after COVID-19 based on an online survey.	June–August 2020	804 participants	Bangladesh	45% expect to cycle more after the COVID-19 pandemic – likely switching from public transport.	Recommendation: Cycling infrastructure is needed to encourage cycling. Bicycle sales taxes should be reduced.
Zhang and Fricker (2021)	Bayesian structural time series model based on trip counter data.	January–May 2020	12 counters	11 cities in the USA	Decreasing walk and bike trips on off-road multi-use trails in densely populated cities, but increases in less densely populated cities.	Not measured.

Our literature review is divided into two main parts – each based on a major theme identified in the literature. First, we review changes in cycling levels during COVID-19 (see column 6 of Table 1 for details). Second, we review government policies during COVID related to cycling: government policies implemented; measured impacts of government policies on cycling levels; and overall recommendations for government policies based on each study’s analysis (see column 7 of Table 1).

2.1. Peer-reviewed studies

To identify peer-reviewed studies, we searched the publication databases of the Web of Science, Scopus, the Transportation Research Board’s TRID (Transportation Research International Documentation) database, and the World Health Organization’s COVID-19 Research Database. We used the following search terms to identify relevant publications published between 1 January 2020 and 25 January 2023: bike, bicycle, bicycling, cycling, active travel, active transport – in combination with Coronavirus, COVID, and COVID-19. We searched the titles, abstracts, and keywords of publications. The initial search found 95 bicycling-related publications in Web of Science, 91 in Scopus, 57 in TRID, and 72 in the WHO database. Eliminating overlap yielded 149 unique studies.

Based on reading the abstracts and scanning the texts of those 149 publications, we excluded 49 that were not relevant to our topic of the possible impact of COVID-19 on ordinary, non-professional cycling. For example, several papers dealt with professional sports cycling. Some excluded articles focused on topics such as air quality during COVID due to changes in travel behaviour but only briefly mentioned cycling without any cycling-specific analysis. We also excluded studies that aggregated walking and cycling, making it impossible to isolate changes specific to cycling. We included such studies, however, if they examined government policies benefiting both walking and cycling, such as reduced speed limits, car-free streets, neighbourhood traffic calming, and off-road shared paths. Finally, a few studies included the terms COVID, COVID-19, or Coronavirus in the abstract or keywords but did not deal at all with the topic, as we discovered upon reading the full text. Eliminating papers not directly relevant to our topic yielded 100 peer-reviewed papers that we included in our review.

The 100 peer-reviewed studies we reviewed were listed in the literature databases between March 2020 and January 2023. Most of the studies examined the time period 2019–2020; 22 studies included data from 2021, and only two of the studies included data from 2022. As a whole, the studies cover a worldwide sample of six continents, 53 countries, and 261 cities. Europe (41) and the USA (29) accounted for the most cases studied, followed by Asia (23), Canada (13), Australia (9), Latin America (8), and Africa (4). Some studies examined several countries on several continents. As reported in Table 1, the peer-reviewed studies (column 1) used a variety of methods and types of data (column 2), time periods (column 3), sample sizes (column 4), and geographies (column 5) to estimate effects of the COVID-19 pandemic on cycling levels (column 6) and government measures to promote cycling (column 7).

2.2. Non-refereed studies and datasets

Using Google Scholar and our own informal inquiries with academic and professional colleagues, government agencies, and cycling organisations, we identified 12 other papers, reports, and datasets that were directly relevant to our study. They are listed in Table 1 with asterisks to distinguish them from the peer-reviewed studies. Our search for non-refereed publications was highly selective, based primarily on filling in gaps and updating information in the peer-reviewed literature. In some cases, these non-refereed sources were broader in scope and more up-to-date than the refereed articles and included larger sample sizes. For example, the Eco-Counter dataset (and its descriptive analysis) was the only source of comprehensive, international information available on cycling levels on a week-by-week basis over the entire period from 1 January 2019 through 31 December 2022.

As listed in Table 1 and the references, most of the studies that were not peer-reviewed were produced by non-governmental organisations, government agencies, and university research centres. Their geographic coverage included Europe, North America, and Australia. Eight of the studies examined the time period 2019–2020. Four of the studies examined the period 2019–2021; two of those examined the entire period 2019–2022. Although their analysis is mostly descriptive, the non-refereed sources provide a useful complement to the refereed publications by reporting more recent information and using larger databases.

3. Changes in cycling levels during COVID

3.1. Measuring cycling levels

Of the 86 studies we reviewed that assessed changes in cycling levels during COVID, the three main data sources were: surveys (43 studies), automatic counters (13), and bikesharing pick-up and drop-off data (12). Less frequently used data sources were interviews (7), GPS-based data (7), and crowd-sourced data from mobile apps or social media (4). Although surveys were most frequently used, there was considerable variation in survey type. Some surveys posed a general question of whether the respondents cycled more frequently during COVID than before COVID. Some surveys asked about the specific frequency of cycling or how many trips were made by bike during COVID compared to before COVID. Similar to the automatic counter data, such surveys generated data about levels of cycling in the time periods compared but no information on the percentage of trips by bike. Several of the studies, however, relied on surveys that were multimodal, asking about trips by all modes over the compared time periods, and thus enabling calculations of the mode share of cycling compared to other modes. Some surveys asked specifically about trip purpose, while others did not. Similarly, surveys varied by the extent to which they asked details about time of day, day of the week, and location of trips as well as socio-demographic details about the respondents. A few studies asked not only about past cycling levels but also future intentions.

Automatic counters were generally able to distinguish between cyclists and pedestrians but provided no information about trip purpose or socioeconomics. Unlike the survey data, they provided detailed information about the exact location and timing of trips. Because the counter data can often be distinguished by type of facility – for example, recreational vs. commutation routes – trip purpose can sometimes be inferred.

Several studies relied on data generated by bikesharing systems, which report volumes of bicycles checked-out and checked-in at specific docking-station locations, including information on day of the week and time of day. Floating (dockless) bikesharing systems report specific locations and time of use based on GPS data. A few studies used GPS-based tracking data to measure changes in cycling levels – typically based on data transmitted from mobile phones (e.g. Streetlight) or fitness trackers (e.g. Strava). Similar to the counter data, GPS traces do not provide information about trip purposes and socioeconomics of the travellers. However, they provide detailed information about the specific trip location and time of day. A few studies analysed Twitter feeds to assess general changes in attitudes towards cycling and cycling levels (e.g. cycling more or less than before COVID).

3.2. Overall cycling trends

Of the 60 peer-reviewed studies analysing changes in cycling from 2019 to 2020, 32 found increases in cycling, while 8 reported decreases, and 20 reported both increases and decreases, with differences by period of the pandemic, geography, day of the week, time of day, trip purpose, and demographics. For example, studies including many cities generally found increases in some cities but decreases in others. In addition, some studies that examined changes over several time periods found decreases over some time periods but increases in others. Of the 7 non-refereed studies focusing on changes between 2019 and 2020, 4 reported increases, while 3 reported both increases and decreases. Of the 14 peer-reviewed studies with data including 2021, 11 found increases in cycling from 2020 to 2021, while 3 studies found both increases and decreases. The two non-refereed studies including data for 2021 and the two non-refereed studies including data for 2021 and 2022 all reported both increases and decreases. Overall, more studies reported increases than decreases, but there was much variation among studies. Moreover, studies that disaggregated by period of the pandemic, geography, day of the week, time of day, trip purpose, and demographics found that changes in cycling levels diverged along those dimensions. The following sections of this paper explore that variation.

3.3. Variation by period of the pandemic

Most studies reported that cycling levels were higher in 2020 than in 2019 – or higher in specific pandemic periods in 2020 than in pre-pandemic periods in 2020 (see Table 1, column 7). Exceptions were studies that compared cycling during complete lockdowns in 2020 to pre-pandemic cycling, all of which found decreases during lockdowns (Echaniz et al., 2021; Heydari et al., 2021; Li et al., 2021; Teixeira et al., 2021). Of the 18 studies (both peer-reviewed and refereed) including data from 2021 and 2022, most found decreases or only slight increases in cycling from 2020 to 2021 or from 2020 to 2022 – but increases from 2019 to 2021 and 2022 (see Table 1, column 7).

Of all the studies examined, only Eco-Counter provides weekly data on cycling levels over the entire period 1 January 2019 to 31 December 2022, thus enabling an examination of detailed trends in cycling levels continuously during that four-year period (Eco-Counter, 2023). By comparison, almost all of the other studies examined only two – or at most three – specific points in time. Although Eco-Counter data are only available for 11 European countries, Canada, and the USA, they provide a unique tracking of the situation week-by-week over four years and thus portray fluctuations throughout the pandemic. For the period 2019–2020, Eco-Counter reported increases in cycling levels averaging 8% in 11 European countries, 16% in the USA, and 3% in Canada (Eco-Counter, 2023). From 2020 to 2021, Eco-Counter reported an average increase of 2% in Europe, no change in Canada, and a decrease of 13% in the USA. Over the entire period from 2019 to 2021, cycling levels increased by an average of 10% in Europe compared to 3% in both the USA and Canada.

In their analysis of the weekly Eco-Counter data, Buehler and Pucher (2022) found that the largest percentage declines in 2020 (relative to 2019) were in periods of full lockdowns with stay-at-home orders, but even less severe restrictions – closed offices, shops, restaurants, schools, and universities – greatly reduced travel overall, including cycling. Similar to other studies, they found large increases in cycling after lockdowns, usually exceeding levels in the same months of 2019 prior to COVID-19. Eco-Counter reports much smaller fluctuations over time in 2021 than in 2020 and a slight decline in overall cycling levels from 2020 to 2021. For the entire four-year period 2019–2022, cycling rose in 10 countries and fell in 3 countries (Eco-Counter, 2023).

3.4. Variation by geography

Several of the studies we reviewed reported variation in cycling trends among countries. Of the 11 European countries, the USA, and Canada in the Eco-Counter database, changes from 2019 to 2021 ranged from overall decreases of 3% in Germany and 7% in Finland to overall increases of 23% in the UK and 27% in Italy (Buehler & Pucher, 2022). Barbieri et al. (2021) collected survey data in May 2020 in Australia, Brazil, China, Ghana, India, Iran, Italy, Norway, South Africa and the USA. Compared to pre-pandemic levels, they found less cycling in May 2020, which was a period of COVID travel restrictions in most of those countries. However, the extent of the declines varied among countries from moderate to negligible. In a review of 43 peer-reviewed journal articles (as of March 2021) on global trends in cycling during the pandemic, Nikitas et al. (2021) reported more increases than decreases in cycling, but with large variation among 42 cities in 22 countries on 5 continents.

Studies also found variation within countries. For example, Eco-Counter (2023) reported variation in cycling changes among geographic regions of the USA over the period 1 January 2019 to 31 December 2022, ranging from −3% in the Southwest, −1% in the Rocky Mountain region and the Southeast, +1% on the Pacific Coast, +8% in the Northeast, to +20% in the Midwest. On the basis of 89 bicycle counters throughout Norway, Nordengen et al. (2021) found a 23% increase in cycling in southern Norway from 2018 to 2020, but an 8% decrease in cycling in northern Norway.

Several studies found large differences among cities in the same country. Rérat et al. (2022) found that from 2019 to 2021, the frequency of cycling among cyclists rose by 44% in Lausanne and by 42% in Geneva. By contrast, cycling levels in Zurich and Basel fell from 2019 to 2020 but rebounded in 2021 to pre-COVID levels (Büchel et al., 2022). A study of ten cities in Germany found that cycling levels rose from 2019 to 2020 in eight cities but fell in two cities (Möllers et al., 2022). In the United States, Streetlight (Morzynski et al., 2020; Musulin et al., 2021) reported cycling increases between July 2019 and July 2021 in 81 out of 100 cities but decreases in 19 cities. In general, cycling increased the most in cities with low cycling levels in 2019 but decreased in cities with high cycling levels, especially in university towns (due to remote learning) and in cities with high levels of bike commuting to work (due to remote working). This finding is consistent with declines in cycling mode share between 2019 and 2021 reported by national travel surveys in The Netherlands (28% to 26% of trips) and Denmark (15% to 12% of trips), Europe’s most bike-oriented countries (CBS, 2022; DTU, 2022). The national travel survey in Japan also reported a fall in the cycling mode share of trips: from 13% in 2015 to 10% in 2021 (MLIT, 2023).

There may have been inherently less potential for increased cycling during COVID in countries and cities that already had high levels of cycling in 2019 (Buehler & Pucher, 2022; Büchel et al., 2022; Möllers et al., 2022). Moreover, in countries and cities with high levels of cycling, most bike trips are utilitarian (Pucher & Buehler, 2008). As many studies reviewed in the next section of this article found, daily trips to work, school, university, and shopping tended to decline during COVID, while recreational cycling increased. In addition, several studies we reviewed noted that cities that already had high cycling mode shares and extensive high-quality cycling networks in 2019 undertook few, if any, special COVID-related measures to increase cycling (Buehler & Pucher, 2022; Büchel et al., 2022; de Haas et al., 2022; Möllers et al., 2022). In combination, those three factors might help explain the decline of cycling in countries and cities that had high cycling mode shares in 2019.

As documented in Buehler and Pucher (2022), some cities that invested massively in expanding their cycling infrastructure during COVID (such as Paris, London, and Brussels) experienced large increases in cycling – except during strict lockdowns. One peer-reviewed study reported a “cycling boom” in Manila and two other metro areas in the Philippines after the national government funded more than 500 km of new cycling facilities specifically as a response to COVID (Sunio & Mateo-Babiano, 2022). Several studies measured increases in cycling on new cycling facilities, especially pop-up bike lanes installed during COVID (Becker et al., 2022; Kraus & Koch, 2021). From 2019 to 2021, cycling levels remained roughly the same in Zurich and Basel (Büchel et al., 2022), which did not implement any special pro-cycling policies, but rose considerably in Lausanne and Geneva, which made large investments in cycling infrastructure (Rérat et al., 2022).

3.5. Variation by trip purpose

Many studies reported large increases in cycling for recreation, exercise, stress reduction, and getting outdoors but reductions in daily commutation trips to work, university, and school (Buehler & Pucher, 2021a; Geiger et al., 2021; Kearns & Wright, 2022; Shibayama et al., 2021). For example, using crowd-sourced data from Strava, Schweizer et al. (2021) found a 55% increase in recreational and exercise cycling in parks of 15 large German cities. Using GPS tracking data and an online survey, Molloy et al. (2021) documented large increases in recreational cycling but a drop in commuter cycling in Switzerland. An international survey with 11,000 respondents reported decreases in bicycle commuting in 9 of 12 countries – with a majority of former bicycle commuters having shifted to working from home in almost all countries (Shibayama et al., 2021).

Some of the growth in cycling was due to an increase in the proportion of round trips – starting and ending at home or on off-road greenways – for recreation, stress relief, or exercise routines (Gladwin & Duncan, 2022; Rérat et al., 2022). Related to that difference by trip purpose, many studies reported especially large increases in cycling in or close to parks on off-road paths such as greenways but decreases on key commutation corridors to city enters, commercial districts, and universities (Fischer et al., 2022; Geiger et al., 2021; Hong et al., 2022).

3.6. Variation by time of day and day of the week

Related to differences by trip purpose, cycling generally increased the most on weekends and weekday afternoons while decreasing during weekday mornings (Chen et al., 2022; Fuller et al., 2021; Geiger et al., 2021; Molloy et al., 2021). In the 11 European countries included by Eco-Counter, cycling increased from 2019 to 2020 by an average of 23% on weekends but by only 3% on weekdays. Over the same period, cycling in the United States increased by 29% on weekends and 10% on weekdays; cycling in Canada increased by 28% on weekends but declined by 8% on weekdays (Eco-Counter, 2023).

In Europe, the USA, and Canada, cycling on weekdays decreased during the morning commute hours, but increased in the mid to late afternoon (Eco-Counter, 2023). Kearns and Wright (2022) examined cycling changes from March 2019 to March 2020 at 70 counter locations throughout the state of North Carolina. They found that the largest increases in cycling were on off-road recreational greenways, especially on weekends and late weekday afternoons and evenings. Similarly, Monfort et al. (2021) found that in Arlington County, Virginia mid-day cycling traffic increased by 76% between 2019 and 2020, but that morning cycling traffic declined by 49%. Nguyen and Pojani (2022) found a large increase in recreational cycling in Hanoi, Vietnam from 2019 to April 2021, but it was concentrated in the early morning (5–8 am) – before heavy motorised traffic later in the day made cycling less safe. Using app-based GPS tracking, Molloy et al. (2021) found that the largest increases in cycling in Switzerland from March-August 2019 to March-August 2020 were on weekends, holidays, and weekday afternoons. They concluded that the increased cycling was for leisure, exercise, and to get outside – not for the commute to work.

3.7. Variation by socio-demographics

Several studies report varying trends in cycling levels during COVID by gender, age, and employment status. For example, relying on a survey of 315 seniors in Teheran, Iran, Shaer and Haghshenas (2021a) found that increased senior cycling in 2020 was limited to men, while women increased their walking during the pandemic. In their study of recreational cycling in Hanoi, Vietnam, Nguyen and Pojani (2022) found that men were more likely than women to start cycling during the COVID pandemic. Using structural equations modelling to investigate cycling in four cities in South America (Buenos Aires, Quito, Santiago, and Lima) in the fall of 2020, Vallejo-Borda et al. (2022) found that female and older public transport riders were less likely to shift to active travel compared to men and younger individuals.

Based on 419 entries in 22 qualitative mobility diaries of women in three large African cities (Abuja, Cape Town, and Tunis) in 2020 and 2021, Porter et al. (2021) recommended cycling and walking as viable alternatives for low-income women who still had to go outside the house during COVID to find food and water and to earn income – in spite of the risk for infection. Using interviews, sketches, and self-generated videos with study participants, Waitt and Stanes (2022) found that lower motorised-traffic volumes in Sydney, Australia during COVID allowed women to start riding bicycles. In a survey of 136 urban areas in 39 countries during COVID, Fischer and Gopal (2021) found that cycling was the main form of physical activity for 20% of men, but for only 8% of women. The same study found that cycling was the main form of physical activity for 18% of respondents in high-income countries compared to only 4% in low-income countries.

Based on a logistic regression of survey data from Philadelphia for 213 respondents in the summer of 2020, Cusack (2021) found a lower likelihood of commuting to work by bicycle for non-whites and for people with special time constraints and concerns about safety. In districts close to Santiago de Chile’s CBD, Valenzuela-Levi et al. (2021) reported significant increases in cycling to work among those who were not working from home (25%–32%). Nguyen and Pojani (2022) found increases in the share of children cycling to school (2%–13%) in Hanoi, Vietnam – mainly replacing walking trips.

3.8. Bikesharing

With few exceptions (e.g. Ku et al., 2021 and Song et al., 2022), studies reported that bikesharing fell sharply at the start of the pandemic due to fear of contagion from handling the same bike as others. It also fell during strict lockdowns during which most travel was prohibited. Remote working replaced most commute trips during lockdowns and has remained at higher levels than in 2019 even after 2020 (Heydari et al., 2021; Jobe & Griffin, 2021; Li et al., 2021). Due to the sharp decline in public transport usage in most countries, bikesharing was also less used than previously for getting to and from public transport stations (Buehler & Pucher, 2022).

Soon after the onset of COVID, many bikesharing systems made a concerted effort to sanitise bikes. Moreover, it eventually became clear that COVID is mainly spread through airborne particles emitted by people, especially indoors. That led to a revival of bikesharing in many cities (Chen et al., 2022; Heydari et al., 2021; Jobe & Griffin, 2021). Studies found that bikesharing during COVID was used more than previously for longer recreational trips instead of work trips or short access trips to and from public transport (Chen et al., 2022; Li et al., 2021; Xin et al., 2022). Sangveraphunsiri et al. (2022) found that bikesharing recovered more quickly from COVID-19 trip reductions than other modes of transport.

3.9. Shift from public transport to cycling

Several studies found a shift from public transport to cycling, not only due to health concerns but also cutbacks in public transport service. In an analysis of 15,776 English-language tweets dealing with transport issues during COVID, Habib and Anik (2021) found that 34% of Twitter users discussed shifts to bicycling from public transport due to health concerns. Möllers et al. (2022) found shifts from public transport to cycling in all 10 German cities they examined. Schaefer et al. (2021) found a shift of trips from local light rail to bicycling in Hannover, Germany. In Bangladesh, Zafri et al. (2021) found that 45% of respondents considered switching to bicycling from public transport.

In a study of bikesharing in Shenzhen, China, Cao and Wang (2022) found that reduced public transport service due to COVID increased demand for bikesharing. Similarly, Heydari et al. (2021) found that public transport riders in London shifted to bikesharing during the pandemic – often resulting in longer bikeshare trips compared to trip duration before the pandemic. Song et al. (2022) concluded that bikeshare served as a reliable alternative when public transport systems were closed in Singapore.

4. COVID and government policies related to cycling

Especially in 2020, many cities throughout the world introduced a range of measures to facilitate cycling, walking, and other uses of streetspace such as outdoor dining or provision of extra space for social distancing. Some of those measures relied on reallocating streetspace from motorised to non-motorised uses and restricting motor vehicle use through reduced speed limits, prohibition of motor vehicle traffic on certain days or times of day, and restrictions on thru traffic in residential neighbourhoods.

4.1. Overall cataloguing of measures

Combs and Pardo (2021) used worldwide data gathered from crowdsourcing, professional networks, and systematic internet searches during a five-month period between 9 March and 9 August 2020. They identified over a thousand specific street space reallocation measures. Of those measures, 13% were shifts of motor vehicle travel lanes to cycling or walking, 22% were full or partial street closures to motorised traffic, and 8% involved bikesharing expansion or bike rental discounts. In a follow-up analysis, Kutela et al. (2022) used text mining and network analysis to examine 483 measures over the longer period March 2020 to January 2022. In both their network analysis and logistic regression, they found that bike lanes were the most likely to be made permanent, followed by full-street closures to motor vehicles (open streets). Overall, Kutela found that 75% of the 483 measures were only short-term and not permanent. Similarly, Glaser and Krizek (2021) found that only 6 of 30 large American cities had made COVID measures permanent.

The COVID-19 Measures Tracker of the European Cyclists Federation (ECF, 2022) found that from March 2020 to September 2022, roughly 2600 km of new bikeway infrastructure, traffic calming, and car-free areas were announced by local, regional, state, and national level governments in Europe. By September 2022, 1500 km of those measures had been implemented, including 43 of 94 of the largest cities (ECF, 2022; ECF & Eco-Counter, 2023). Roughly 1.7 billion Euros have been allocated for bicycling in Europe since the start of the pandemic – in the form of additional funding for infrastructure, tax incentives/reductions for cycling, subsidies for bicycle purchases and repair, and promotional campaigns (ECF, 2022; ECF & Eco-Counter, 2023). During 2020, approximately 200 U.S. cities reallocated street space to increase outdoor activity, including cycling (PfB, 2021).

4.2. Bikeway infrastructure

Many of the studies in Table 1 report the expansion and improvement of cycling infrastructure in cities throughout the world (see column 7). Brisbane, Melbourne, and Sydney all expanded their cycling facilities, mainly pop-up bike lanes (Infrastructure Australia, 2020). As reported by Lin et al. (2021), Toronto greatly expanded its network of low-stress cycling facilities in response to COVID in 2020. Within 4 months of the onset of COVID, 106 European cities installed an average of 11.5 km of pop-up bike lanes per city (Kraus & Koch, 2021). As a specific response to COVID, the Philippine national government allocated USD 23 million for a 500 km bike lane network in the country’s three major metropolitan areas, including Metro Manila, where 313 km were already completed by 30 June 2021 (Sunio & Mateo-Babiano, 2022). In Geneva and Lausanne, Switzerland, the pandemic accelerated implementation of existing cycling plans, with each city building 8 km of new cycle lanes in Spring 2020. The space for the new lanes was reallocated from car traffic lanes in Geneva, but from car parking lanes in Lausanne (Rérat et al., 2022).

All 14 large European and North American cities examined by Buehler and Pucher (2022) expanded and improved their cycling infrastructure, mainly by building protected bike lanes (physically separated from motor vehicle traffic), some of which started as pop-up bike lanes but were made permanent because they were successful in attracting more cycling. Some of the new protected bike lanes, however, had already been included in previously existing bike plans, with their installation accelerated during COVID (e.g. in London, New York, and Montreal). In addition, most of 14 cities examined in the study undertook bicycle lane widening, buffering, and engineering design improvements, including intersection modifications to increase cycling safety and convenience. In order to facilitate the expansion and improvement of their cycling networks, all 14 of the cities increased their funding for cycling and increased engineering and planning staff dealing with cycling infrastructure and programmes.

Off-road facilities in both urban and rural areas expanded in the USA. For example, there was a 21% increase in the total length of off-road, shared-use paths for cyclists and pedestrians during the three years between 31 December 2019 and 31 December 2022, from 55,427 km to 67,076 km (RTC, 2023). In many cases of infrastructure expansion during COVID, it is not possible to identify how much of the expansion was specifically attributable to COVID. In some cases, however, governments announced the measures as being explicitly in response to COVID – such as the 500 km of new bike lanes in the Philippines and the Coronapistes (Corona bike lanes) in Paris.

Several studies we reviewed specifically examined the relationship between new cycling infrastructure and cycling levels. Based on data from 736 bicycle counters in 106 European cities, cycling increased on pop-up bike lanes in mid to late 2020 by 11%–48% compared to cycling on the same streets before COVID (Kraus & Koch, 2021). Detailed analysis of six pop-up cycling facilities in Sydney found large increases in cycling, ranging from 30% to 500% growth between July 2020 and April 2021. Because total travel declined over that period, bike mode share rose by even higher percentages (Harris & McCue, 2023). After installation of a pop-up bike lane in Berlin, cycling increased by 73% along that street from 2019 to 2020 (Becker et al., 2022). Based on Strava data, Fischer et al. (2022) found that the largest increases in cycling in Vancouver from 2019 to 2020 were on “comfortable riding spaces” such as protected on-street bike lanes and bikeways in and near parks. Similarly, Hong et al. (2022) found that cycling in Glasgow, Scotland increased the most on protected bikeways during COVID.

Of the 14 European and North American cities examined by Buehler and Pucher (2022), 13 expanded the supply of bike parking in 2020 and 2021. Some cities also increased its quality by providing shelter, improved racks, and security at key locations. It is not clear, however, to what extent such bike parking improvements were explicitly a response to COVID. Nor did any studies we reviewed measure the extent to which improved bike parking during COVID might have raised cycling levels.

In contrast to studies examining actual changes in cycling, simulation studies used statistical modelling to estimate probable changes. For example, simulations for Trieste, Italy estimated that cycling would have increased considerably during COVID if there had been better cycling infrastructure (Scorrano & Danielis, 2021). In a simulation study of Patna, India, Thombre and Agarwal (2021) estimated that building a bicycle superhighway would increase the bike share of trips to work from 31% to 44%. These simulation studies are hypothetical, of course, but we include them in our review because they attempt to isolate the possible impacts that specific infrastructure improvements would have had on cycling levels during COVID.

4.3. Bikesharing

Many bikesharing programmes suspended operations during the height of the pandemic in their cities. A study by the US Bureau of Transportation Statistics (BTS) found that from March to December 2020, 51 bikesharing systems in the USA ended operations permanently (mainly dockless bikesharing systems), 23 suspended operations temporarily, and 62 remained open (BTS, 2022). Many local governments took measures to restore the attractiveness of bikesharing. For example, some cities made special efforts to sanitise their bikes, while others offered temporarily discounted or even free bikesharing (Combs & Pardo, 2021).

Of the 9 European and 5 North American cities examined by Buehler and Pucher (2022), several cities, such as New York, Vancouver, Portland, and Barcelona, expanded bikesharing during COVID and included more E-bikes in their fleet (Buehler & Pucher, 2022). In 2020, Bergantino et al. (2021) surveyed 1163 respondents in Italian cities (including 766 from cities without bikesharing) and found that many more residents of those cities would ride bikes if bikesharing existed or would be expanded.

4.4. Restrictions on car use

Many large European and North American cities implemented road closures and speed reductions (Buehler & Pucher, 2022). Combs and Pardo (2021) identified roughly 250 full or partial street closures in response to COVID in cities around the world. Glaser and Krizek (2021) found in June 2020 that in 30 large U.S. cities, the most common COVID-related roadway measures were experiments with “open streets” or “slow streets”, usually prohibiting thru traffic and lowering speed limits. In a 6-month follow-up (December 2020), they found that 50 percent of the programmes (15 cities) were still in effect, including 6 cities that had expanded them (30 percent) and made permanent some experiments. Landgrave-Serrano and Stoker (2023) found that Slow Streets more than doubled cycling and walking in Tucson, Arizona compared to control streets – in spite of the hot climate and lack of shade.

In their review of 9 European and 5 North American cities, Buehler and Pucher (2022) found that most of the cities imposed new or more extensive restraints on car use during COVID. Most cities reduced speed limits, either city-wide, on designated slow streets, or in low-traffic, traffic-calmed neighbourhoods. For example, Paris and Brussels reduced their city-wide limits on most streets to 30 km/h, Washington, DC and Portland to 20 mph (32 km/h), and New York City to 25 mph (40 km/h). Brussels banned motorised thru-traffic from the city centre, reducing traffic volumes by a third (Buehler & Pucher, 2022). In only six months in 2020, London implemented additional Low Traffic Neighborhoods (LTNs) in areas of the city housing 300,000 residents. The primary intent of LTNs is to discourage thru motor vehicle traffic by installing various traffic calming devices to reduce traffic volumes, but traffic calming also tends to reduce vehicle speeds (Aldred et al., 2021). LTNs make walking, cycling, and playing in streets safer and less stressful.

New York City implemented the largest speed camera programme in the Americas, with 2200 cameras operating in 750 school speed zones. Those cameras reduced speeding by 72 percent in the school districts where they were installed (Buehler & Pucher, 2022). From 2020 to 2022, London closed 420 streets near schools to motorised traffic during morning drop-off and afternoon pick-up times for students (Thomas et al., 2022). Vancouver, Canada also implemented car-free streets adjacent to some schools during school hours (Buehler & Pucher, 2022).

Some cities banned motor vehicle traffic on certain streets at certain times of day, in effect creating car-free streets. In a study of 314 US cities, Evenson et al. (2023) divided their city sample into two parts. They found that 51% of the larger cities (>165,000 residents) implemented car-restrictive measures compared to only 16% of small cities (<165,000). In 2020, New York City banned motorised traffic from 133 km of residential streets. In one of the most widely publicised measures, Paris transformed Rue de Rivoli, a major arterial street in central Paris, into a bicycle street that allows motor vehicle traffic in only one lane in one direction to enable bike traffic in the remaining 4 lanes in both directions. Portland permanently reduced the width of many of its arterial roads from 5 lanes to 3 lanes, freeing up space for cyclists and reducing car speeds. Most cities at least temporarily removed car parking or traffic lanes on portions of some streets in 2020 to enable more room for outdoor dining and to provide space for pop-up bike lanes (Buehler & Pucher, 2022; Combs & Pardo, 2021; Evenson et al., 2023). Barcelona, for example, removed 21 km of traffic lanes to provide space for expanded bike lanes.

Car drivers tended to oppose car-restrictive measures. In Berlin, for example, pedestrians (75%), cyclists (94%), and public transport users (79%) were supportive of pop-up bike lanes, while car drivers opposed them (79%) (Becker et al., 2022). Similarly, a 2020 survey of 2011 people in the USA found 34%–39% public support for car restrictions or roadway reallocations to promote safer bicycling and walking in June 2020 – in contrast to 59%–64% support (almost twice as much) for mask mandates and rules requiring physical distancing in public spaces (Duren et al., 2021).

4.5. Equity

Several studies examined the social equity impacts of cycling-related government policies implemented during COVID. Fischer and Winters (2021) compared the socio-spatial equity impacts of street reallocation measures in three Canadian cities intended to encourage cycling, walking, and social distancing. The measured social equity results were mixed. In both Victoria and Kelowna (both in British Columbia), street reallocations tended to be more concentrated in areas of lower-income and higher proportions of Black and Indigenous peoples. In Halifax, Nova Scotia, street reallocations were slightly less concentrated in minority areas than for the city as a whole but neutrally distributed by income. In all three cities, the planning process for street reallocations was participatory and included equity considerations. Firth et al. (2021) examined the equity impacts of street reallocations favouring cyclists and pedestrians in Vancouver (40 km) and Seattle (20 miles). They found the most streetspace reallocations in neighbourhoods with high proportions of residents who were people of colour, particularly Black and Indigenous peoples. The authors emphasise the need for community involvement in street reallocation decisions. Using GIS to analyse 59,544 network links, Lin et al. (2021) found that the expanded low-stress cycling infrastructure implemented in Toronto, Canada during COVID greatly increased accessibility by bike to essential destinations, including areas of the city with low-income and minority populations.

In their global review of COVID-era transport measures, Combs and Pardo (2021) report social equity problems in some of the more than a thousand measures they catalogued from around the world, but without any mention of the number of cases where this was a problem. Their main concern appears to be what they call “top-down decision-making and lack of robust participatory components in the planning and implementation process”, which caused “vigorous backlash” in some instances. They emphasise that a crucial aspect of the equity impacts of COVID-era measures is that they were embedded in a long-standing pattern of transport, housing, land use, and law enforcement policies that systematically excluded racial and ethnic minorities and low-income groups from access to safe active mobility for many decades prior to COVID (Martens et al., 2021).

Aldred et al. (2021) found that Low Traffic Neighborhood (LTN) implementation in London during COVID has been broadly equitable. Across London as a whole, residents of the most deprived neighbourhoods were 2.5 times more likely to live in a new LTN compared to residents of the least deprived neighbourhoods. Black, Asian and Minority Ethnic (BAME) people were slightly more likely than White Londoners to live in a new LTN. Transport for London, which implements LTNs, explicitly uses equity criteria related to deprivation in its planning processes. Thomas et al. (2022) found that School Streets in London have been equally distributed across several socio-demographic indicators. Fischer et al. (2022) used Strava data to track changes in cycling levels and socio-demographics in Vancouver during COVID. They found that cycling increased most on low-stress, comfortable facilities such as protected bike lanes, slow streets, and off-road paths – which expanded during COVID. Increased cycling by women and older adults on such facilities narrowed somewhat the age and gender gap among cyclists. In their analysis of Santiago de Chile, Valenzuela-Levi et al. (2021) found that subsidised housing within cycling distance to job concentrations helped increase equity in job access during COVID, when lower-income workers still needed to travel to work. Thus, the authors advocate a post-COVID situation with more affordable housing built near job concentrations so that trip distances are short enough to be covered by bike.

Our review found that during COVID, many cities throughout the world have focused on providing protected bike lanes and other low-stress facilities such as slow streets, closed streets, and traffic-calmed neighbourhood streets. Previous research found that low-stress cycling facilities are crucial to increasing cycling by children (McDonald et al., 2021), older adults (Garrard et al., 2021), women (Garrard, 2021), and by anyone who is especially sensitive to cycling safety (Furth, 2021). Thus, the focus on expanding such low-stress, safer cycling facilities especially targets the needs of children, older adults, and women, and has a positive equity impact along those demographic dimensions (Buehler & Pucher, 2021b). In a study of Toronto, Canada, Hassen (2022) concluded that expanding networks of protected bikeways into lower-income neighbourhoods would improve the health of their residents.

In a study of the COVID response of bikesharing systems in the USA, Tiako and Stokes (2021) found that many bikesharing systems offered free or low-cost memberships to essential workers who had to commute to work in spite of COVID. Those workers were more likely to be Black or LatinX. In theory, that would seem to be an equitable policy. In practice, however, inclusion criteria of these programmes were often arbitrary – for example, excluding food and hospitality workers in New York, San Francisco, and Boston. Thus, the discounts only partially benefited lower-income and minority workers and were hardly sufficient to overcome the striking inequity in both the distribution of docking stations and the affordability of bikesharing in most American cities (Fishman & Shaheen, 2021; Martens et al., 2021; Tiako & Stokes, 2021).

4.6. Policy recommendations of reviewed studies

In addition to the simulation studies noted previously, almost all studies examined conclude by recommending expansion and improvements to cycling infrastructure (Abdullah et al., 2022; Advani et al., 2021; Buehler & Pucher, 2022; Ehsani et al., 2021; Irawan et al., 2022; Scorrano & Danielis, 2021; Shaer & Haghshenas, 2021a; Thombre & Agarwal, 2021). Several studies also recommend permanent and more widespread implementation of complementary policies such as traffic calming, slow streets, slower city-wide speed limits, and prohibiting non-local traffic in residential neighbourhoods (Buehler & Pucher, 2022; ECF, 2022; Landgrave-Serrano & Stoker, 2023; Shirgaokar et al., 2021).

As a whole, the reviewed studies suggest a wide range of justifications for promoting cycling: increased mobility options for all segments of the population, especially for lower-income groups; improved environmental quality; improved traffic safety; reductions in greenhouse gas emissions; and health benefits of increased physical activity.

Studies found that local government support was crucial not only for funding but also for implementing potentially controversial policies such as reallocating roadway space from motor vehicles to bicycles (Buehler & Pucher, 2022; ECF, 2022; Glaser & Krizek, 2021; Nikitas et al., 2021). Many pro-cycling and car-restrictive policies thought impossible before the pandemic became possible to implement due to the public and political support generated by a crisis situation (Buehler & Pucher, 2022; Combs & Pardo, 2021; Harris & McCue, 2023; PBF, 2021, 2022).

Several studies emphasised the importance of community involvement and widespread stakeholder participation in decision-making about measures to promote cycling and walking, both during COVID as well as longer term (Aldred et al., 2021; Combs & Pardo, 2021; Firth et al., 2021; Hassen, 2022; Valenzuela-Levi et al., 2021). Many studies examining equity impacts of COVID-era measures found that the most equitable results were in cases where the planning authorities had specific guidelines to ensure consideration of the needs of disadvantaged groups, especially racial and ethnic minorities and low-income groups. As noted by several studies, it is also important to conduct studies of the actual equity impact of measures after they have been implemented to ensure they have had the intended equitable impact.

5. Limitations and gaps in the existing literature

An inevitable limitation of all the studies we reviewed is that none of them can prove a causal link between COVID and cycling – at best a statistically significant relationship in studies using multivariate analysis. Correspondingly, our review of the literature cannot conclude that COVID was necessarily the cause of the observed cycling trends. Studies examining new or expanded government policies related to cycling were not always able to show that COVID was the only or main reason for implementing these policies. As noted earlier, however, there are exceptions to this, such as instances when a government explicitly introduced a new measure as a response to COVID. In addition, most studies were not able to determine to what extent government policies implemented during COVID were the reason for observed changes in cycling levels. Some before-and-after studies of specific new facilities (such as pop-up bike lanes) suggest a causal relationship, but they do not control for other factors that might have changed at the same time.

The studies we reviewed used different data sources and methods. Some relied on data from automatic counters while others reported results from travel surveys, in-depth interviews, bikesharing GPS data, social media such as Twitter, or riding apps such as Strava. There are potential problems of data reliability, variable definition and measurement, and methodology for all of these sources. Studies relying on surveys reported issues of bias in the selection of participants, bias and possible inaccuracy of self-reported data, limited timing (usually two points or periods in time), limited geographic coverage, and small sample sizes. Some studies relying on automatic counters reported limited numbers of such counters and the lack of statistically representative locations of counters. Studies relying on data from Strava, Streetlight, and Twitter users are also biased to the extent that such users are not representative of the population as a whole.

In addition to such limitations of data and methods, the studies we reviewed are not directly comparable. They examine a wide range of cities of different types and sizes in different parts of the world. Due to the different timing of COVID and the resulting lockdowns and travel restrictions among and within countries, the studies cover different time periods. Most of the studies compare 2019 with 2020, but several extend the comparison to 2021 and 2022. Some studies compare entire years, while others compare specific periods in each year. Only the Eco-Counter data provide week-by-week data on cycling levels over the entire 4-year period 1 January 2019 to 31 December 2022, but they are limited to the USA, Canada, and 11 countries in Europe and thus not representative of other parts of the world.

Future research should consider the long-term impacts of COVID-19. Most of the studies in this review rely on data from 2019 and 2020, with only a few extending into 2021, and even fewer extending into 2022. It would be useful to have follow-up studies in a few years to determine the longer-term impacts of COVID-19 on bicycling. Because the COVID pandemic has been so recent, all of the studies that we reviewed were forced to rely on short-term data, and some of the studies relied on non-representative or small samples. That probably limits the reliability of the estimated impacts.

There is also a need for more studies of the social equity implications of government policies to promote cycling, and more socioeconomic analysis of trends in cycling over time (Martens et al., 2021). As reported by several of the studies we reviewed, there has already been encouraging movement in this direction, but more needs to be done to ensure that social equity considerations are explicitly incorporated in all government transport policies. In addition, there is a pressing need for more studies of cycling levels and policies in the Global South, which was underrepresented in the studies we reviewed.

6. Conclusions

The literature we reviewed – published between March 2020 and January 2023 – does not confirm the worldwide bike boom reported by much of the media in 2020, the first year of the pandemic. Overall, studies published so far suggest more increases than decreases in cycling from 2019 to 2020, with some cities reporting large increases. Nevertheless, there has been much variation among countries, cities, and specific corridors within cities as well as variation by gender, age, ethnicity, and income group. There have also been ups and downs in cycling levels corresponding to waves of COVID and the resulting lockdowns and restrictions on travel, which have differed in timing and severity from one place to another, even within countries. The largest increases in cycling in 2020 were for recreation, exercise, and stress relief on weekends and weekday afternoons. By comparison, cycling to work, university, schools, and shopping generally declined.

Most of the cities examined by the studies we reviewed reported expansions or improvements in their bikeway network, often specifically related to COVID or accelerated due to COVID. In many cases, the new facilities were implemented on a trial basis but were so well-used that they were made permanent. Rising cycling levels also provided an incentive for cities to plan for even greater improvements to bikeway networks in the coming years (Buehler & Pucher, 2022). In addition, many of the studies we reviewed concluded by recommending more investment in protected cycling infrastructure, in particular.

Importantly, most cities also implemented new or expanded restrictions on motor vehicle parking and use, especially the reallocation of roadway space to cyclists, pedestrians, and outdoor eating. Whether in the form of lane reallocation, car-free roads and spaces, speed reductions, or elimination of thru traffic in residential neighbourhoods, such restrictions often faced motorist opposition. Given the emergency circumstances of COVID, however, even cities in extremely car-oriented countries were able to implement car-restrictive measures. Some were only temporary and have since been rescinded or curtailed, but their implementation made it possible to test the potential of such measures to encourage more cycling.

As emphasised above in the section on policy recommendations of reviewed studies, one important lesson of COVID-19 is that it generated public and political support for large increases in pro-cycling policies combined with restraints on motor vehicle use that would have been considered impossible under normal circumstances. The experience with a wide range of new or expanded measures during COVID revealed the effectiveness of low-stress, safe facilities such as protected bike lanes, off-road paths, slow streets, school streets, and traffic-calmed neighbourhoods to encourage a wider range of the population to ride bikes. Finally, the pandemic demonstrated the importance of bicycling as a sustainable means of transport that not only can complement other modes but also provide a feasible alternative to them when crisis situations interrupt their normal operation, thus contributing to the resilience of the overall transport system.

Disclosure statement

No potential conflict of interest was reported by the author(s).