The prognostic significance of exercise testing was compared with clinical and electrocardiographic (ECG) variables in a prospective study of 107 patients with unstable angina discharged from the hospital on medical therapy. During a follow-up period of 12.8 +/- 1.4 months, 10 patients (9%) had a nonfatal myocardial infarction (n = 8) or died (n = 2) and 22 (20%) were readmitted with recurrent unstable angina. The relation between 20 clinical, ECG and exercise test variables and the risk of adverse outcome (death, nonfatal myocardial infarction or recurrent unstable angina) was analyzed using both univariate and multivariate (logistic regression) analysis. Univariate predictors of adverse outcome included diabetes mellitus, evolutionary T wave changes, T wave changes on the preexercise ECG and low maximal rate-pressure product during exercise. Independent predictors of adverse outcome in multivariate analysis included diabetes mellitus, evolutionary T wave changes after admission, rest pain during hospitalization, ST depression during exercise and low maximal rate-pressure product. A predictive model constructed using the regression equation and all independent predictors stratified patients into high and low risk groups (41% and 5% risk of adverse outcome, respectively). The result of a predischarge exercise test adds independent prognostic information to clinical and ECG data in medically treated patients with unstable angina and could be used in combination with clinical and ECG data to identify patients at risk of adverse events.