Background: Diabetes mellitus (DM) has been linked to poor prognosis in pancreatic cancer in numerous retrospective datasets, potentially due to higher levels of insulin and insulin-like growth factor increasing cellular proliferation. Additionally, the hyperglycemic state as well as obesity may lead to a pro-inflammatory state that may enhance tumor progression and metastases. In this analysis, the prognostic significances of diabetes, insulin use, and BMI are evaluated in a prospective, randomized trial of resectable pancreatic ductal adenocarcinoma. Methods: As previously reported, RTOG 9704 pts with resected pancreatic cancer were randomized to 5-fluorouracil (5-FU) or gemcitabine (Gem), given pre and post radiation (RT). Pts in both arms received RT to 50.4 Gy with 5-FU. DM and insulin use were collected on the RTOG 9704 CRFs. BMI was dichotomized as normal/underweight vs. overweight/obese. Overall survival (OS) and disease-free survival (DFS) were estimated by Kaplan-Meier method and variable levels were compared using log-rank test. Cox models were used for multivariable analyses. Results: 538 pts were enrolled from 1998-2002. 238 pts were eligible with analyzable diabetes and insulin use data, in addition to surgical pathology, and CA19-9 data. 32% and 35% of pts in the 5-FU and Gem arms, respectively, had DM. Of which, 24% and 19% used insulin in the 5-FU and Gem arms, respectively. 55% were overweight/obese. The 4 yr OS was 25% (95% CI: 18, 32) in pts with no diabetes, vs. 18% (10, 27) with DM (HR = 1.3 [0.9, 1.7]; p = 0.11). The 4-yr DFS was 14% (9, 20) in pts with no diabetes, vs. 9% (4, 16) with diabetes (HR = 1.2 [0.9, 1.5]; p = 0.31). Neither insulin use nor BMI was univariately associated with OS or DFS. On multivariate analysis, including nodal status, CA19-9, and other variables, DM, insulin use, and BMI were not associated with OS or DFS. Conclusions: DM, insulin use, and BMI were not associated with OS in pts with resected pancreatic cancer treated with adjuvant chemotherapy and chemoradiation. Node involvement and CA19-9 remain the most significant predictors of OS. Supported by NCI grants U10CA180868, U10CA180822, UG1CA189867 and Eli Lilly Clinical trial information: NCT00003216