Three defibrillators, one manual and two different semiautomatic, were prospectively compared during a one year period for out-of-hospital use by ordinary ambulance personnel with short additional training. Eighty-three cardiac arrest patients were treated with one of two different semiautomatic defibrillators and 26 by an ordinary manual defibrillator. Twenty-nine were found in ventricular fibrillation. Five of these patients were successfully resuscitated and admitted for further hospital care, two survived to be discharged home. The semiautomatic defibrillators were found to be accurate in ECG interpretation, sensitivity and specificity was 100%, respectively, and both were equally effective in defibrillation. There were no differences in conversion rate or in the clinical outcome between the three defibrillators tested. Both semiautomatic defibrillators tested seemed to be safe, reliable and cost-effective. The low survival rate found is most certainly due to a long ambulance delay.