Many studies have shown improved survival of cardiac arrest patients by the use of early defibrillation (EMT-D) in the field. This prospective study was the first in Pennsylvania and was undertaken to determine if an EMT-D program would be successful in our suburban/rural setting. One hundred two EMTs were trained to use a semi-automatic defibrillator and data were collected over 16 months. There were 96 cardiac arrests, with only 33 patients (34%) presenting with initially treatable dysrhythmias-ventricular fibrillation (VF) or tachycardia (VT). Twenty-three patients (24%) were admitted to the hospital; survival to hospital discharge occurred in only 5 patients (5.2%). Survival to hospital admission was higher among VF/VT presenting rhythms (36%) than for those with other rhythms (17%, P = 0.07), but survival to discharge among VF/VT rhythms (9%) was not statistically different from other rhythms (3%, P = 0.45). Among VF/VT patients, survival to discharge was correlated with shorter call to first defibrillation intervals. Mean call to response interval was longer than in other reported studies (7.2 ± 4.3 minutes). In addition, there was a high drop-out rate of EMT participants, no central/uniform early access system (that is, 911), and a lower rate of CPR than reported in other studies. It is concluded that introduction of an EMT-D program without careful analysis of systems response factors will not lead to the improved cardiac arrest survival percentages that have previously been reported. © 1993.