We studied survival rates among 767 men with good left ventricular function who participated in the European Coronary Surgery Study, 10 to 12 years after they were randomly assigned to either early coronary bypass surgery or medical therapy. At the projected five-year follow-up interval, we observed a significantly higher survival rate (+/- 95 percent confidence interval) in the group that was assigned to surgical treatment than in the group assigned to medical treatment (92.4 +/- 2.7 vs. 83.1 +/- 3.9 percent; P = 0.0001). During the subsequent seven years, the percentage of patients who survived decreased more rapidly in the surgically treated than in the medically treated group (70.6 +/- 5.8 vs. 66.7 +/- 5.3 percent at 12 years). Thus, the improvement in the survival rate among patients with stable angina who were treated surgically appears to have been attenuated after five years. However, the gradually diminishing difference between the two survival curves still favored surgical treatment after 12 years (P = 0.04), despite the fact that 136 patients in the medically treated group had coronary bypass surgery and 23 in the "surgically treated" group did not. The benefit of surgical treatment tended to be greater, but not significantly so, as assessed by interaction analysis in the subgroups of patients who were older or who had signs of ischemia or previous infarction on the resting electrocardiogram, a markedly ischemic response to exercise testing, peripheral arterial disease, an absence of hypertension, and proximal obstruction in the left anterior descending artery. The reasons for the loss of a beneficial effect of surgery after five years are unknown and merit further study.