A cohort of 317 diabetic patients, aged > or = 65 years, with angiographically proven coronary artery disease, was analyzed and followed for a mean of 12.8 years. Compared with 1,843 age-matched nondiabetic patients, diabetic patients were more likely to (1) have a higher number of coronary occlusions, (2) not be current smokers, (3) have higher systolic but lower diastolic blood pressures, (4) have evidence of peripheral vascular disease, and (5) be women. They did not differ significantly with respect to total cholesterol, family history of coronary artery disease, history of hypertension, or left ventricular hypertrophy. In the total elderly cohort, diabetes was found to be an independent predictor of mortality, conferring a 57.0% increased risk of death. Survival analysis showed that diabetic subjects consistently had higher mortality than nondiabetics. However, the relative survival benefit of coronary artery bypass graft surgery versus medical therapy was comparable in diabetic and nondiabetic patients. Surgical therapy conferred a reduction in mortality of 44%.