To determine if the occurrence and the consequences of spontaneous predischarge postinfarction ischemia could be predicted early after hospital admission, a consecutive series of patients with acute myocardial infarction was studied and followed for 3 years. No patient was treated by thrombolysis. Spontaneous predischarge ischemia was defined as angina that occurred at rest before hospital discharge, at least 3 days after the acute event, and that was accompanied by electrocardiographic changes, but not by an increase in cardiac enzymes. Patients who died within the first 3 days were excluded from analysis. Among the 943 patients who survived at least 3 days, 165 (17.5%) had spontaneous ischemia before discharge. They had a higher 1-year post-hospital mortality (16 vs. 10%), but did not have significantly higher total 3-year mortality rates. Four independent, early available variables predictive of the occurrence of spontaneous ischemia were selected from a stepwise logistic discriminant analysis: history of angina before infarction, non-Q-wave infarct, absence of smoking, and higher age. Among the 165 patients with spontaneous ischemia, 3 independent variables predictive of 3-year mortality were selected stepwise: left ventricular function score, history of previous infarction, and absence of smoking.