PURPOSE: Subtrochanteric femoral shaft fractures after little or no trauma have been reported in long-term users of bisphosphonates, but risks relative to hip fracture protective effects and among men are not clear. We examined associations between bisphosphonate use and nontraumatic subtrochanteric (NTST) femoral fractures and hip fractures in the Veterans Health Administration. METHODS: This retrospective cohort study was conducted using 1998-2007 Veterans Health Administration electronic medical records data on 78,155 individuals who had a fragility fracture at age 45 years or older. Time-to-event analysis examined associations of bisphosphonates with risk of NTST femoral fracture and, separately, hip fracture, controlling for sociodemographics, medications, and comorbid medical conditions. RESULTS: The cohort had a mean age 66.5 years (32.5% were ≥75 years old) at the time of their first fracture, and 69.3% were observed for 6 or more years; only 11.8% were prescribed bisphosphonates during observation. During follow-up, 408 had an NTST femoral second fracture, and 1584 had a hip second fracture. Compared with those never on bisphosphonates, the adjusted hazard ratio for NTST femoral second fracture among patients on 4 years of therapy or longer was 0.40 (95% confidence interval, 0.16-0.97) and for hip second fracture was 0.38 (95% confidence interval, 0.24-0.61). CONCLUSIONS: Bisphosphonate treatment in this high-risk cohort was infrequent with few long-term users, limiting power to assess long-term effects. Nontraumatic subtrochanteric femoral fractures were uncommon, and longer bisphosphonate use was associated with lower (not higher) risk. In men, risks of NTST femoral fractures associated with bisphosphonate treatment may be low in contrast to substantial protective benefits for hip fracture.