To determine the long-term influence of the severity of preoperative diabetes mellitus on the results of coronary bypass, a review was made of 212 diabetics operated on between 1968 and 1973, of whom 87 patients (41%) were receiving no drugs, 108 patients (50.9%) were receiving oral hypoglycemic agents, and 17 patients (8%) were receiving insulin. They were compared with 1,222 nondiabetic patients operated on over the same period. Perioperative mortality was similar in the diabetics and nondiabetics: 7.1% vs 4.5%. Improvement in anginal symptoms was similar in all patient groups: 85.9% to 92.7%. Overall 15-year survival probability was .53 for the nondiabetic group, .43 for the diabetics not receiving drugs, .33 for those receiving oral agents, and .19 for the insulin-treated patients. Late graft patency ranged from 78% to 90% and was comparable in all groups. The preoperative blood glucose level was an important predictor of late mortality in all diabetic patients. Thus, coronary bypass surgery was effective in all groups of diabetic patients in long-term relief of anginal symptoms. Intermediate-term survival rates were good in all groups, but the initial severity of the diabetes was an important determinant of long-term survival rates.