Summary It remains unclear whether the association between metformin and osteoporosis (OP) risk is causal. This two-sample Mendelian randomization (MR) study suggests a causal relationship between metformin treatment and a decrease in OP and fracture incidence, as well as an increase in bone mineral density (BMD) in the lumbar spine, femoral neck, and heel. Nonetheless, no significant causal effect is observed on forearm BMD. Purpose We utilize a MR approach to investigate the association between metformin treatment and the risk of OP. Methods Metformin treatment was selected as exposures. Outcomes included OP; BMD at the forearm (FA), femoral neck (FN), and lumbar spine (LS); estimated heel bone mineral density (eBMD); and fracture. Summary statistics for exposures and outcomes were obtained from corresponding genome-wide association studies. Inverse variance-weighted (IVW) analysis was mainly applied; the weighted median (WM), penalized weighted median (PWM), maximum likelihood (ML), and MR-Egger regression (MR-Egger) method were also used to obtain robust estimates. A series of sensitivity analyses including Cochran’s Q test, MR-Egger regression, leave-one-out analysis, and Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO) were used to detect pleiotropy or heterogeneity. Results In the main analysis, IVW estimates demonstrated that metformin treatment had a definite causal effect on the risk of OP (odds ratio (OR): 0.859, 95% CI: 0.774–0.953, P = 0.004), LS-BMD (OR: 1.063, 95% CI: 1.023–1.105, P = 0.002), FN-BMD (OR: 1.034, 95% CI: 1.000–1.069, P = 0.049), eBMD (OR: 1.035, 95% CI: 1.023–1.047, P ≤ 0.001), and fracture(OR: 0.958, 95% CI: 0.928–0.989, P = 0.008). However, it did not have an effect on FA-BMD(OR: 1.050, 95% CI: 0.969–1.138, P = 0.237). Conclusions This study indicated that metformin treatment is significantly associated with a reduction in the risk of OP, fracture and higher LS-BMD, FN-BMD, and eBMD. However, there was no significant association with FA-BMD.