Published reports of out-of-hospital cardiac arrest give widely varying results. The variation in survival rates within each type of system is due, in part, to variation in definitions. To determine other reasons for differences in survival rates, we reviewed published studies conducted from 1967 to 1988 on 39 emergency medical services programs from 29 different locations. These programs could be grouped into five types of prehospital systems based on the personnel who deliver CPR, defibrillation, medications, and endotracheal intubation; the five systems were three types of single-response systems (basic emergency medical technician [EMT], EMT-defibrillation [EMT-D], and paramedic) and two double-response systems (EMT/paramedic and EMT-D/paramedic). Reported discharge rates ranged from 2% to 25% for all cardiac rhythms and from 3% to 33% for ventricular fibrillation. The lowest survival rates occurred in single-response systems and the highest rates in double-response systems, although there was considerable variation within each type of system. Hypothetical survival curves suggest that the ability to resuscitate is a function of time, type, and sequence of therapy. Survival appears to be highest in double-response systems because CPR is started early. We speculate that early CPR permits definitive procedures, including defibrillation, medications, and intubation, to be more effective.