OBJECTIVES: Several predictors of type 2 diabetes mellitus (T2DM) remission after metabolic surgery have been proposed and used to develop predictive scores. These scores may not be reproducible in diverse geographic regions with different baseline characteristics. This study aimed to identify predictive factors associated with T2DM remission after Roux-en-Y gastric bypass (RYGB) in patients with severe obesity. We hypothesized that the body composition alterations induced by bariatric surgery could also contribute to diabetes remission. METHODS: We retrospectively evaluated 100 patients with severe obesity and T2DM who underwent RYGB between 2014 and 2016 for preoperative factors (age, diabetes duration, insulin use, HbA1c, C-peptide plasma level, and basal insulinemia) to identify predictors of T2DM remission (glycemia<126 mg/dL and/or HbA1c<6.5%) at 3 years postoperatively. The potential preoperative predictors were prospectively applied to 20 other patients with obesity and T2DM who underwent RYGB for validation. In addition, 81 patients with severe obesity (33 with T2DM) underwent body composition evaluations by bioelectrical impedance analysis (InBody 770®) 1 year after RYGB for comparison of body composition changes between patients with and those without T2DM. RESULTS: The retrospective analysis identified only a C-peptide level >3 ng/dL as a positive predictor of 3-year postoperative diabetes remission, which was validated in the prospective phase. There was a significant difference in the postoperative body composition changes between non-diabetic and diabetic patients only in trunk mass. CONCLUSION: Preoperative C-peptide levels can be useful for predicting T2DM remission after RYGB. Trunk mass is the most important difference in postoperative body composition changes between non-diabetic and diabetic patients.