The predictive value of a postoperative exercise test in terms of cardiac events after coronary artery bypass grafting (CABG) was prospectively studied in 231 consecutive patients. During a 5-year follow-up there were 28 cardiac events (12%), of which 15 were cardiac deaths (13 sudden), and 13 were nonfatal myocardial infarctions. There was no difference in the rate of graft patency between groups with and without cardiac events, but ejection fraction was lower in patients with than without events (51 +/- 16% vs 58 +/- 10%; p less than 0.05). Duration of the exercise test was shorter, and maximal work load was lower in patients with cardiac events (p less than 0.05 for both). The prevalence of greater than or equal to 1 mm ST-segment depression was 22% (symptomatic in 25%, and silent in 75%) and did not differ between groups with and without cardiac events. After adjustment for prognostic variables using the proportional hazards method, diuretic treatment (p = 0.007) and a low postoperative ejection fraction (p = 0.04) remained significant for predicting the risk of cardiac events within 5 years of CABG, but exercise duration and work load did not have any significant predictive value. Thus, the predictive value of a postoperative exercise test is limited, and signs of impaired left ventricular function are of greater significance for the 5-year prognosis after CABG than are those of myocardial ischemia.