Background: The antidiabetic drug M may improve cancer-specific survival; however, little is known about the association between M use and survival among NSCLC pts. Methods: We retrospectively identified pts diagnosed with stage I-IV NSCLC from the Veterans Affairs Central Cancer Registry. A diagnosis of type 2 diabetes (DM) within 6 months of diagnosis was obtained from VA administrative data. Anti-DM medication (med) use, including M, was identified using VA pharmacy data. We used Cox proportional hazards regression controlling for age, race, BMI, stage, co-morbidities, HbA1C, and NSCLC treatment to estimate adjusted hazard ratios (AHR) to compare mortality risk among (1) DM pts with M use (2) DM pts with non-M anti-DM med use (3) DM pts with no anti-DM med use and (4) non-DM pts. Results: Between 2001-2008, 51,824 pts were diagnosed with NSCLC, 19% (n=9,982) of whom had a diagnosis of DM. Of the DM pts, 33% (n=3,309) were treated with M, 41.2% (n=4,113) were treated with non-M anti-DM meds, and 26% (n=2,560) were not treated with any anti-DM meds. When compared to non-DM pts, DM pts treated with M had similar survival [AHR: 0.99; 95% CI: (0.96, 1.03)] while DM patients treated with non-M anti-DM meds [AHR: 1.14; 95% CI: (1.10, 1.18)] or who received no anti-DM meds [AHR: 1.10; 95% CI: (1.05, 1.15)] were associated with poorer survival. DM pts treated with M were associated with improved survival compared to DM pts treated with non-M anti-DM [AHR: 0.86; 95% CI: (0.82, 0.91)]. DM pts treated with M had a similar improved survival relative to DM pts treated with non-M anti-DM meds across all stage groups [stage I/II: AHR 0.85; 95% CI: (0.77-0.93), stage III: AHR 0.88; 95% CI: (0.81-0.97), stage IV: AHR 0.86; 95% CI: (0.80-0.99)], including stage IA [AHR 0.83; 95% CI: (0.75-0.93)]. Conclusions: M use in NSCLC pts with DM is associated with (1) similar survival to non-DM pts, and (2) improved survival among DM NSCLC pts. These associations were observed regardless of stage, suggesting a broad anti-NSCLC effect of M or an unidentified confounding factor.