We show that automated external defibrillation training of emergency medical technicians (EMTs) is less time consuming than manual defibrillation training, and hypothesize that both improve survival from sudden cardiac death. Data on 91 cardiac arrests over 27 months among five basic life support services was collected before EMT-defibrillation (EMT-D) training. Subsequently, seven BLS services were trained in EMT-D using either manual defibrillation or automated external defibrillation technology, and 55 sudden cardiac death patients were entered after training. Manual defibrillation required 11 more hours per student in initial training. Survival to hospital discharge improved from two of 91 patients (2.2%) in the series before EMT-D training to nine of 55 patients (16.4%) after EMT-D training (P = .001). Improved survival was correlated with shorter prehospital defibrillation times, 8.84 minutes, when EMTs performed defibrillation versus 16.3 minutes before training when EMTs awalted advanced life support defibrillation (P < .001). To enhance equipment famillarity we allowed EMTs to apply three-lead electrode monitors to all medical/cardiac patients during transport (surveillance). There were six emergency medical service-witnessed "surveillance" arrests and three arrests survived to hospital discharge (50% survival). This group represented 33% of all survivors in the series. We recommend automated external defibrillation training for EMTs. Improved survival in sudden cardiac death cases in well-run emergency medical service systems should result from EMT-D training. Finally, we recommend that routine "surveillance" of high-risk patients during transport by defibrillation-capable EMTs be considered in EMT-D programs, rather than limiting EMT-D only to units capable of rapid "man-down" response. © 1993.