The degree of left ventricular impairment in an acknowledged important prognostic marker of long-term outcome for patients being evaluated for implantation of cardioverter-defibrillators. Just how left ventricular function impacts freedom from all-cause mortality, as well as from sudden death and cardiac death, is a subject of current major debate, and is analyzed hereunder from a large, recent multicenter ICD patient cohort. The multicenter database consists of data from 361 patients receiving implantable cardioverter-defibrillators for standard indications, that is, documented episodes of ventricular fibrillation or sustained ventricular tachycardias with poor hemodynamic toleration. Data were collected from 1988 to 1995 at three centers in Germany. Two-hundred and three patients (56%) had a left ventricular ejection fraction (LVEF) > 0.30 (group I), and 158 patients (44%) had a LVEF < or = 0.30 respectively (group II). The mean follow-up was 23.9 months (range 3-98 months). Overall survival at 5 years for group II patients was lower, as expected, at 74.1% versus 94.2%, respectively (P < 0.0001). Mortality was higher for each different cause of death in group II patients than in Group I: sudden arrhythmic deaths, 5 versus 1 (P < 0.048); nonsudden cardiac deaths, 16 versus 5 (P < 0.002); noncardiac deaths, 7 versus 2 (P < 0.03). Group II patients received a higher rate of at least one presumably appropriate shock at 86 (54.4%) versus 89 (43.8%) in group I (P < 0.05). However (and somewhat surprisingly), neither the time from ICD implantation to death, comparing only the patients who died, nor the event-free probability of appropriate shocks due to very rapid, sustained ventricular arrhythmias (> 230 beats/min), including a presumed risk of sudden arrhythmogenic death, differed between groups I and II. Sudden cardiac death was only marginally affected by LVEF (group I, 1.5% actuarial, 5-year survival 99.5%; group II, 3.1% and 95.8%, respectively). Therefore, the lower overall survival in ICD patients with LVEF < or = 0.30 resulted mainly from causes of death that cannot be directly influenced by cardioverter-defibrillator therapy. However, because group II patients had a far higher incidence of at least one ventricular tachyarrhythmia terminated by ICD shocks than group I patients, they also probably derived benefit from ICD therapy.