In order to further define clinical and angiocardiographic predictors of long-term survival after myocardial infarction we followed 616 consecutive male patients under 60 years of age, survivors of a first (N = 455) or recurrent (N = 161) myocardial infarction, for 8.8 +/- 2.9 years. Patients had angiocardiography at 4-8 weeks after infarction; none had thrombolysis, but 33% had cardiac surgery, 14% on a clinical trial basis. Left ventricular end-systolic volume was the most powerful predictor of cardiac mortality; ejection fraction and end-diastolic volume added no further information. Myocardial score, a measure of the severity of coronary stenoses in relation to the amount of myocardium supplied, was of only borderline predictive value on multivariate analysis, possibly because any effect had been negated by coronary surgery. Administration of beta-blocker drugs had an independent effect of improving prognosis, while continued cigarette smoking worsened it. Age, status of index infarction (first or recurrent) and serum cholesterol did not affect survival. A trial of surgery, carried out in a subset of 200 of these patients who were relatively asymptomatic but had severe coronary disease, showed no survival advantage for intended surgical over non-surgical management. We conclude that a high left ventricular end-systolic volume remains the most important adverse prognostic factor after recovery from myocardial infarction.