Abstract
Ventricular fibrillation is shown to persist for 12 or more minutes in victims of cardiac arrest, while survival rates me known to decline steadily, with time delay from cardiac arrest to advanced cardiac care. These two facts imply that VF alone is not an appropriate condition upon which t o base community or non-randomized comparisons of emergency medical services. A minimal data set necessary for such comparisons is suggested and two models for making the comparisons are presented.