To determine the prognostic significance of ventricular arrhythmias persisting during the hospital ambulatory phase of acute myocardial infarction, 64 patients with acute myocardial infarction underwent continuous 10-hour Holter monitoring an average of 11 days after discharge from the coronary care unit (CCU). Patients were categorized according to the results of ambulatory monitoring: 27 patients had ventricular extrasystoles, which were complicated (multifocal, R on T, paired, more than 5/min), or ventricular tachycardia; 22 had uncomplicated premature ventricular contractions; and 15 exhibited no ventricular arrhythmias. The 64 patients were followed prospectively for an average course of 25.8 months; 12 died suddenly; 8 died of other causes, and 44 survived. In all patients who died suddenly, ventricular ectopy was recorded on Holter monitoring before their discharge from the hospital (complicated premature ventricular contractions, eight patients; uncomplicated premature ventricular contractions, four patients); there were no sudden deaths in the patients without ventricular arrhythmias. Patients who died suddenly and those survived were similar in respect to age (60, 62 years), sex, location of infarction, presence of coronary risk factors, severity of acute myocardial infarction (Q waves, cardiac enzymes), serum cholesterol levels, evidence of cardiomegaly on roentgenograms, presence of ventricular gallop and drug therapy received. The occurrence of acute arrhythmias in the CCU did not separate patients who died suddenly from those who survived; there were no differences in ventricular tachycardia or ventricular fibrillation (3 or 12 patients who died suddenly, 6 of 44 patients who survived) or complicated premature ventricular contractions (4 or 12 patients who died suddenly, 18 of 44 patients who survived). Electrocardiograms obtained late in the hospital course revealed no differences in the extent of Q or T wave changes between these two groups. However, the extent of S-T segment abnormality was greater in patients who died suddenly than in patients who survived (5.6 compared to 1.8 leads/standard tracing, p smaller than 0.02) suggesting that the arrhythmias in the former were related to persistent ischemia or segmental ventricular dyssynergy. Thus, in this relatively small number of patients, ventricular arrhythmias persisting late in the hospital course of patients admitted for acute myocardial infarction are shown to predispose to subsequent sudden death.