Internal defibrillation leads were placed at time of coronary revascularization in 79 patients. In 34, an implantable cardioverter defibrillator (ICD) was placed simultaneously (group I). A two-stage strategy (selective implantation of the ICD in patients with postoperative spontaneous or inducible ventricular tachycardia [VT]) was followed in 45 patients (group II). Group I patients had failed more antiarrhythmic drug trials (2.9 +/- 1.6 vs 1.5 +/- 1.6; P = 0.02), including amiodarone (62% vs 20%; P < 0.001). There were four operative deaths in each group. Postoperatively, VT was present in 27 group II patients (60%), 25 of whom received an ICD (two refused device implantation). Patients with postoperative VT had a lower left ventricular ejection fraction than those without VT (33 +/- 9 vs 47 +/- 16; P = 0.01). Actuarial survival at 1, 2, and 3 years was 88 +/- 6, 88 +/- 7, and 88 +/- 10 in group I; and 83 +/- 6, 76 +/- 7, and 76 +/- 11 in group II (NS). No patient without an ICD (based on the postoperative electrophysiological study [EPS]) died suddenly. Five patients (6%) had ICD system infection. Sudden death was largely prevented by either strategy, but relatively high rates of operative mortality and ICD system infection were observed. Prospective studies should identify patients more likely to benefit from one or another strategy.