We evaluated long-term survival after coronary-artery bypass grafting in 686 patients with stable angina who were randomly assigned to medical or surgical treatment at 13 hospitals and followed for an average of 11.2 years. For all patients and for the 595 without left main coronary-artery disease, cumulative survival did not differ significantly at 11 years according to treatment. The 7-year survival rates for all patients were 70 per cent with medical treatment and 77 per cent with surgery (P = 0.043), and the 11-year rates were 57 and 58 per cent, respectively. For patients without left main coronary-artery disease, the 7-year rates were 72 and 77 per cent in medically and surgically treated patients, respectively (P = 0.267), and the 11-year rates were 58 per cent in both groups. A statistically significant difference in survival suggesting a benefit from surgical treatment was found in patients without left main coronary-artery disease who were subdivided into high-risk subgroups defined angiographically, clinically, or by a combination of angiographic and clinical factors: (1) high angiographic risk (three-vessel disease and impaired left ventricular function)--at 7 years, 52 per cent in medically treated patients versus 76 per cent in surgically treated patients (P = 0.002); at 11 years, 38 and 50 per cent, respectively (P = 0.026); (2) clinically defined high risk (at least two of the following: resting ST depression, history of myocardial infarction, or history of hypertension)--at 7 years, 52 per cent in the medical group versus 72 per cent in the surgical group (P = 0.003); at 11 years, 36 versus 49 per cent, respectively (P = 0.015); and (3) combined angiographic and clinical high risk--at 7 years, 36 per cent in the medical group versus 76 per cent in the surgical group (P = 0.002); at 11 years, 24 versus 54 per cent, respectively (P = 0.005). Survival among patients with impaired left ventricular function differed significantly at 7 years (63 per cent in the medical group versus 74 per cent in the surgical group [P = 0.049]) but not at 11 years (49 versus 53 per cent). The surgical treatment policy resulted in a nonsignificant survival disadvantage throughout the 11 years in subgroups with normal left ventricular function, low angiographic risk, and low clinical risk, and a statistically significant disadvantage at 11 years in patients with two-vessel disease.(ABSTRACT TRUNCATED AT 400 WORDS)