During a 10-year period, 5631 cardiac arrests were treated in our paramedic system. In all, 4216 resuscitations were attempted, of which 533 (12.6%) resulted in saves, defined as hospital discharges. Patients presenting with an initial rhythm of coarse ventricular fibrillation or ventricular tachycardia were found to have significantly increased save rates in comparison to those presenting with an initial arrest rhythm of asystole/fine ventricular fibrillation or electromechanical dissociation (P less than or equal to 0.01). When controlling for witnessed arrest, 303 of 1905 (15.9%) of all witnessed arrests were saves vs. 230 of 2311 (10%) of unwitnessed arrests (P less than or equal to 0.01). Witnessed bystander/first responder external cardiac compression- cardiopulmonary resuscitation (ECC-CPR) was found not to influence save rate. One hundred eighty-one of 1248 bystander/first responder witnessed arrests (14.5%) who had external ECC-CPR initiated before paramedic advanced life support arrival were saves, compared with 38 of 252 (15.1%) who had no ECC-CPR initiated until paramedic arrival; this was not statistically significant. Advanced life support response times in saved patients with witnessed cardiac arrests were analyzed. Ninety-five percent of all saves had a response time of less than 10 min. We conclude that, when evaluating the effectiveness of CPR, the variables of witnessing of arrest, presenting arrest rhythm, and respective response times must be controlled or analyzed.