Clinical factors and terminal events associated with sudden death in 51 patients were analyzed from among a multicenter experience of 864 recipients of first generation automatic implantable cardioverter defibrillator (AICD) devices (single zone, committed, monophasic pulse with > or = 1 epicardial patch electrode) during the period May 1982-February 1988. For these 51 patients, mean age was 58 years and atherosclerotic heart disease was present in 84%, with a history of ventricular fibrillation (VF) in 61%, and inducible sustained ventricular tachycardia (VT) in 84%; mean left ventricular ejection fraction was 0.26. Nearly 80% experienced one or more appropriate AICD shocks during the median 9 month (range 0-46 months) period prior to death. Of 30 monitored deaths, the first documented terminal rhythm was VF in 12 (40%), VT in 8 (27%), and asystole or electromechanical dissociation in the remaining 10 (33%). Shocks were documented during terminal events in 21 (66%) of 32 witnessed cases of sudden death with activated devices. The proportion of monitored or witnessed sudden deaths that were known or presumed to be tachyarrhythmic (based on terminal VT, VF, or shocks) ranged from 69% (11/16 cases with activated/nondepleted devices and a defibrillation threshold [DFT] < or = 20 J) to 81% (29/36 cases on intention-to-treat basis). Of 27 patients with known or presumed sudden tachyarrhythmic death, the AICD had been deactivated prior to death in 4 (15%); activated, but depleted in 4 (15%); activated/nondepleted, but with DFT of 25 J in 4 (15%); and activated/nondepleted, but without DFT testing in 4 (15%). The remaining 11 (41%) known or presumed sudden tachyarrhythmic deaths occurred in patients with activated/nondepleted devices and DFT < or = 20 J; however, definite or suspected contributory factors (e.g., hematoma under epicardial patch, generator component failure, or drug-induced DFT rise) could be identified in 6 (55%) of 11 cases. Thus, in this first-generation AICD experience: 1) most sudden deaths occurred on the basis of a known or presumed tachyarrhythmia; and 2) an understanding of apparent "failure" of ICD therapy could often be gained through an integrated analysis of associated clinical factors and management practices, as well as device "hardware" function. These observations are likely to remain relevant, even with respect to newer generation ICDs.