To identify predictors of late mortality, 259 consecutive men (less than or equal to 60 years old) who survived acute myocardial infarctions were catheterized one month after admission and were then followed for a mean of 34 months. Nineteen patients (7 per cent) died during the observation period. Of 79 base-line descriptors, 17 proved to be univariate predictors of survival. Cox regression analysis demonstrated that the ejection fraction (P less than 0.001), the number of diseased vessels (P less than 0.005), and the occurrence of congestive heart failure in the coronary unit (P less than 0.01) were the only independent predictors of survival. Risk stratification showed that the probability of survival at four years was highest in patients with normal ejection fractions (96 to 100 per cent, depending on the number of diseased vessels) and lowest in those with ejection fractions below 20 per cent (3o to 75 per cent). The prognosis in patients with ejection fractions between 21 and 49 per cent was significantly worse (78 per cent) than in those with normal ejection fractions only in the group with three-vessel involvement (P less than 0.01). Since most survivors of myocardial infarction who are likely to have their lives prolonged by coronary-artery bypass surgery are in this group, it is reasonable to limit routine coronary angiography to the 56 per cent of survivors who have ejection fractions between 21 and 49 per cent.