Background and aims: To determine the impact of gastric bypass on life expectancy for severely obese patients with diabetes. Materials and methods: We developed a decision-analytic Markov model to evaluate two strategies for severely obese patients with diabetes: gastric bypass versus nonsurgical treatment. The efficacy of surgery was determined from a retrospective cohort of 165,000 severely obese diabetic patients (4500 had gastric bypass) from 4 HMO Research Network sites using fully adjusted Cox proportional hazards models. Logistic regression models calculated in-hospital mortality for surgery using data from the Nationwide Inpatient Sample (NIS). The decision model was calibrated using data from the National Health Interview Survey that is linked to the National Death Index. The model was constructed using Decision Maker®, which estimated changes in life expectancy. Results: Our base case, a 40 year-old female with a BMI of 45, no hypertension, no coronary artery disease and no congestive heart failure, gained an additional 7.1 years of life expectancy with gastric bypass (43.1 years with surgery vs. 36.0 without). Surgery was no longer favored in our base case when 30-day surgical mortality exceeded 16% (baseline risk was 0.2%). Sensitivity analyses revealed that the gain in life expectancy decreases with increasing BMI, until a BMI of 62 is reached; at this point, nonsurgical treatment is associated with greater life expectancy than gastric bypass. Similar results for both men and women in all age groups were seen. Conclusion: For most severely obese diabetic patients, gastric bypass surgery increases life expectancy; however, gastric bypass decreases life expectancy for those with a BMI over 62. Patients with high BMI may have other benefits from surgery, such as better quality of life and reduced burden of comorbid disease.