To determine whether exercise testing can identify patients whose survival might be prolonged by coronary artery bypass surgery, the results of bypass surgery were compared with those of medical therapy alone in 5,303 nonrandomized patients from the Coronary Artery Surgery Study registry who underwent exercise testing. Patients in the two treatment groups differed substantially with regard to important baseline variables. Analysis of 32 variables by Cox's regression model for survival revealed an independent beneficial effect of bypass surgery on survival (p less than 0.00001). Patients were then stratified into subsets according to the results of exercise testing. Surgical benefit was greatest in the 789 patients who exhibited at least 1 mm of ST segment depression and who could exercise only into stage 1 or less. Among the 398 patients with three vessel coronary disease showing these characteristics, 7 year survival was 58% for the medical group and 81% for the surgical group (p less than 0.001). There was no difference in survival between the surgical and medical groups among the 1,545 patients without ischemic ST segment depression who were able to exercise into stage 3 or greater. Thus, in patients who demonstrate ischemia on exercise testing and whose exercise capacity is limited, coronary bypass surgery appears to improve survival in comparison with medical therapy alone.