The prognostic information of clinical variables and a predischarge exercise test was studied in 400 patients (282 men, 118 women) admitted to the coronary care unit with suspected unstable coronary artery disease, that is, recurring chest pain of new onset, increasing anginal pain in formerly stable angina pectoris or suspected nontransmural acute myocardial infarction. Forty-nine coronary events occurred in the 276 men who performed the exercise test during the following year, whereas only 5 coronary events occurred among the 118 women. The only variable of prognostic importance in women was nontransmural myocardial infarction. In men, the clinical variables increasing age, duration of angina, ST- or T-segment changes on the rest electrocardiogram and increasing angina or nontransmural myocardial infarction as inclusion criteria were associated with increased occurrence of coronary artery bypass surgery, transmural myocardial infarction or cardiac death. Findings of ST-segment depression, limiting chest pain or low rate-pressure product during the exercise test were of greater value than any clinical variable in prediction of coronary artery bypass surgery, transmural myocardial infarction or cardiac death. Within all clinical subgroups of men, the results of the exercise test had an additive predictive value for future coronary events. Combinations of clinical data and exercise test results enabled the best identification of patients with high or low risk for coronary events.