Of 232 morbidly obese patients with non-insulin-dependent diabetes mellitus referred to East Carolina University between March 5, 1979, and January 1, 1994, 154 had a Roux-en-Y gastric bypass operation and 78 did not undergo surgery because of personal preference or their insurance company"s refusal to pay for the procedure. The surgical and the nonoperative (control) groups were comparable in terms of age, weight, body mass index, sex, and percentage with hypertension. The two groups were compared retrospectively to determine differences in survival and the need for medical management of their diabetes. Mean length of follow-up was 9 years in the surgical group and 6.2 years in the control group. The mean glucose levels in the surgical group fell from 187 mg/dl preoperatively and remained less than 140 mg/dl for up to 10 years of follow-up. The percentage of control subjects being treated with oral hypoglycemics or insulin increased from 56.4% at initial contact to 87.5% at last contact (P = 0.0003), whereas the percentage of surgical patients requiring medical management fell from 31.8% preoperatively to 8.6% at last contact (P = 0.0001). The mortality rate in the control group was 28% compared to 9% in the surgical group (including perioperative deaths). For every year of follow-up, patients in the control group had a 4.5% chance of dying vs. a 1.0% chance for those in the surgical group. The improvement in the mortality rate in the surgical group was primarily due to a decrease in the number of cardiovascular deaths.