OBJECTIVES: Understanding the benefit-risk profiles of medications for type 2 diabetes (T2D) is of high public health interest. We assessed the potential comparative health benefits and risks of exenatide once weekly (ExQW) compared to insulin glargine (IG) under various assumptions about hypothetical risks from pancreatic and thyroid cancer associated with either treatment. METHODS:  Safety and efficacy data from the randomized controlled DURATION-3 trial were incorporated into the CORE Diabetes Model to predict long-term health outcomes; pancreatic and thyroid cancer risks from SEER (general population); i3/MarketScan claims databases (T2D and/or therapy) were incorporated into derived Markov model.  Incremental net health benefit (INHB) was estimated in life years (LYs) and quality-adjusted life years (QALYs) (both discounted at 3%) for different time horizons (base case=30 years) and different assumptions about cancer risks. The incidence rate of pancreatic cancer was 1.17/10,000 population/year. In the base-case scenario no increased pancreatic cancer risk from ExQW and relative risk (RR) of 1.25 for IG was assumed. The risk for thyroid cancer was assumed same as in general population and equal for ExQW and IG. In two alternative scenarios for pancreatic cancer risk: a) assuming IG RR=1.33 and ExQW having no increased risk, and b) assuming ExQW RR=1.07 and IG RR=1.25.  RESULTS: The INHB for ExQW versus IG was 0.118 Lys and 0.239 QALYs in basecase; 0.119 LYs and 0.239 QALYs in scenario a); and 0.118 LYs and 0.238 QALYs in scenario b).  Neither scenario a) or b) changed INHB by more than1%. The INHB for ExQW remained positive under different assumptions about time horizon and risk levels.   CONCLUSIONS:  In all scenarios modeled, the INHB for ExQW was positive compared to IG.  This study suggests that the potential benefits from ExQW in reducing established T2D complications are expected to outweigh hypothetical risks from pancreatic and thyroid cancer.