The effectiveness of bystander CPR recently has been challenged. We undertook a ten-year retrospective review of our prehospital experience with witnessed cardiorespiratory arrest to ascertain save rates in patients receiving and not receiving CPR before paramedic advanced life support (ALS). Traumatic and poisoning arrests and children less than 18 years old were excluded. A total of 1,905 patients presenting to a paramedic system from November 1, 1973, to October 31, 1983, were bystander-witnessed arrests and attempted paramedic resuscitations. Four hundred five paramedic-witnessed arrests were excluded. One hundred eighty-two of 1,248 (14.6%) who had CPR initiated before paramedic ALS arrival were saves, compared to 38 of 252 (15%) who had no CPR initiated until paramedic arrival (P = NS). A save was defined as a patient discharged from the hospital. The respective save rates for coarse ventricular fibrillation were 148 of 628 (23.6%) (CPR before paramedic arrival) vs 35 of 151 (CPR delayed until paramedic arrival) (23.2%); electromechanical dissociation (EMD), 11 of 209 (5.3%) vs 0 of 38; asystole, 19 of 401 (4.7%) vs 3 of 61 (4.9%); and ventricular tachycardia, four of ten (40%) vs 0 of two. In this prehospital system, bystander/first responder CPR was found not to improve hospital discharge rates except in patients with initially documented rhythm of EMD.