In order to determine those factors which influence long-term prognosis in patients with angina at rest associated with transient ST-segment changes, 217 patients undergoing medical treatment were followed for a mean of 39 months. All patients underwent coronary arteriography. Univariate analysis identified 12 variables significantly related to prognosis. These were disease of the left main coronary artery; the number of diseased vessels; left ventricular end-diastolic pressure; ejection fraction; baseline electrocardiogram; presence of prior myocardial infarction; ST-segment depression and ventricular arrhythmias during pain; disease of the proximal anterior descending coronary artery; crescendo angina; hypertension; and age. Use of the Cox regression model for survival analysis revealed only 3 variables which were independent predictors of prognosis. They were disease of the left main coronary artery; the number of diseased vessels and left ventricular end-diastolic pressure. The model allowed stratification of patients into 3 groups. Survival at 3 years was 98% in the low risk group; 82% in the intermediate risk group; and 58% in the high risk group. These data indicate that disease of the left main coronary artery, the number of diseased vessels and left ventricular end-diastolic pressure are the independent predictors of prognosis in angina at rest. These variables may allow stratification of patients into groups having different long-term survivals.