Until recently, it remains unclear whether schizophrenia, bipolar disorder (BD), and Alzheimer's disease (AD) is associated with bone mineral density (BMD). We aimed to investigate the causal effects of schizophrenia, BD and AD on BMD with Mendelian randomization (MR) analysis. Single-nucleotide polymorphisms (SNPs) strongly associated with these three neuropsychiatric diseases as instrumental variables were selected from genome-wide association studies in the MR Base database. We analyzed the effects of these SNPs on the femoral neck BMD (FN-BMD), lumbar spine BMD (LS-BMD) and forearm BMD (FA-BMD), and evaluated the heterogeneities and pleiotropy of these genetic variants. We also evaluated the potential confounding factors in the association between these three neuropsychiatric diseases and the BMD level. It was found that none of these genetic variants were significantly associated with BMD or confounding factors. Using these genetic variants, we did not find statistically significant causal effects of per unit increase in the log-odds of having schizophrenia, BD or AD with FN-BMD, LS-BMD and FA-BMD changes (e.g. schizophrenia and FN-BMD, MR-Egger OR 0.9673, 95% CI 0.8382 to 1.1163, p = 0.6519). The MR results also revealed that directional pleiotropy was unlikely to bias the causality (e.g., schizophrenia and FN-BMD, intercept = 0.0023, p = 0.6887), and no evidence of heterogeneity was found between the genetic variants (e.g., schizophrenia and FN-BMD, MR-Egger Q = 46.1502, I 2 = 0.0899, p = 0.3047). Our MR study did not support causal effects of increased risk of schizophrenia, BD and AD status with BMD level.