OBJECTIVES: This study evaluated the effect of metformin use on survival in pancreatic cancer patients with curative resection and type 2 diabetes mellitus (T2DM). METHODS: A total of 28,862 were initially identified from Korea Center Cancer Registry (KCCR) who had diagnostic code for pancreatic carcinoma between 1 January 2005 and 31 December 2011. Among them, those with curative resection or T2DM and aged over 40 years were included. Subjects were classified as metformin user group if they were prescribed metformin around the time of diagnosis for pancreatic cancer. Medication possession ratio (MPR) of more than 80% was considered as acceptable adherence. Survival from pancreatic carcinoma was identified from the linkage of National Population Registry of the Korea National Statistical Office with KCCR through December 31, 2013. Several sensitivity analyses were performed to examine the effect of immortal time bias, and confounding variables. RESULTS: The study included 764 subjects with T2D and pancreatic cancer with curative resection, 530 of which were exposed to metformin. In multivariable analysis, the adjusted risk for pancreatic cancer specific mortality of metformin user was significantly lower than that of metformin non-user (hazard ratio, 0.73; 95% CI, 0.61 to 0.87; P < 0.001). The adjusted risk for mortality was also significantly lower in MPR of more than 80% compared with that of MPR of less than 80% (HR, 0.60, 95% CI: 0.47-0.76, p-value<0.001). In addition, similar results were found from a serial of seneitivity analysis. CONCLUSIONS: Metformin use in diabetic patients with pancreatic cancer is associated with improved survival. This may provide a rationale for further prospective study of the use of metformin as an adjunct to the standard of care in the treatment of pancreatic cancer.