BACKGROUND: Conflicting evidence remains in the association of testosterone therapy (TTh) with prostate cancer (PCa). This inconsistency maybe due, in part, to the small sample sizes from previous studies and an incomplete assessment of comorbidities, particularly diabetes. OBJECTIVE: We investigated the association of PCa with TTh (injection or gel) and different TTh doses, and determined whether this association varies by presence of diabetes at baseline in a large, nationally representative, commercially insured cohort. DESIGN: We conducted a retrospective cohort study of 189,491 men aged 40-60 years old in the IBM MarketScan® Commercial Database, which included 1,424 PCa cases diagnosed from 2011 to 2014. TTh was defined using CPT codes from inpatient and outpatient, and NDC codes from pharmacy claims. Multivariable adjusted Cox proportional hazards models were used to compute hazard ratios for patients with incident PCa. RESULTS: We found a 33% reduced association of PCa after comparing the highest category (>12) of TTh injections with the lowest (1-2 injections) category (HR = 0.67, 95% CI: 0.54-0.82). Similar statistical significant inverse association for PCa was observed for men who received TTh topical gels (>330 versus 1-60 days' supply). Among non-diabetics we found significant inverse association between TTh (injection and gel) and PCa, but a weak interaction between TTh injections and diabetes (P = 0.05). Conclusion Overall, increased use of TTh is inversely associated with PCa and this remained significant only among non-diabetics. These findings warrant further investigation in large randomized placebo-controlled trials to infer any health benefit by TTh.