Signal-averaged electrocardiography and 24-hour ambulatory electrocardiographic monitoring were performed in 121 elderly patients > 6 months after acute myocardial infarction. All patients had asymptomatic complex ventricular arrhythmias and a left ventricular ejection fraction > or = 40%. Rates of sudden, cardiac, and total death were compared between groups with and without nonsustained ventricular tachycardia and between normal and abnormal signal-averaged electrocardiographic studies. The prevalence of an abnormal signal-averaged electrocardiographic study was 36%. Thirty-seven percent of the patients had nonsustained ventricular tachycardia, and the remaining patients had complex ventricular arrhythmias other than ventricular tachycardia. There were 27 sudden and 48 total cardiac deaths, and 66 deaths from all causes during a mean follow-up period of 30 months. Kaplan-Meier survival analysis showed a lower rate of sudden and cardiac death in the group without nonsustained ventricular tachycardia. Although there was a trend toward a lower rate of sudden death in patients with a normal signal-averaged electrocardiogram, there was no statistical difference in the rates of sudden, total cardiac, or total death between patients with normal or abnormal studies. The negative predictive value of having neither an abnormal signal-averaged electrocardiogram nor nonsustained ventricular tachycardia was 94% for sudden death. In elderly patients with complex ventricular arrhythmias and ejection fraction > or = 40% at least 6 months after an acute myocardial infarction, presence of nonsustained ventricular tachycardia predicted a higher rate of sudden and cardiac death. Signal-averaged electrocardiography alone was not predictive.