The Beta-Blocker Heart Attack Trial was a placebo-controlled, randomized, double-blind clinical trial of the long-term administration of propranolol hydrochloride to patients who had had at least one myocardial infarction. Among 3,837 patients followed up for an average of 25 months, 3,290 (85.7%) had 24 hour ambulatory electrocardiograms performed at the baseline examination. Four classifications of arrhythmia were examined. One of these, the presence of complex ventricular arrhythmias (at least 10 ventricular premature beats/h, or at least one pair or run of ventricular premature beats or multiform ventricular premature beats) was the subgroup of major interest. Regardless of the classification, the presence of arrhythmia identifies a group of patients with a higher risk of total mortality, coronary heart disease mortality, sudden cardiac death and instantaneous cardiac death. The a priori subgroup hypothesis that sudden death would be preferentially reduced by propranolol in patients with complex ventricular arrhythmias was not supported. The relative benefit of propranolol in reducing sudden death for this subgroup was 28 versus 16% for the subgroup without ventricular arrhythmia (relative risk of 0.72 versus 0.84, a nonsignificant relative difference of 14%). There were similar findings for two of the three other classifications of arrhythmia and for the other response variables. Although propranolol does not appear to be of special relative benefit in patients with ventricular arrhythmia, the presence of the arrhythmia does identify a high-risk group. The mechanism by which propranolol reduces mortality is still unclear, but is probably not solely an antiarrhythmic one.