Abstract Objectives Evidence from observational studies suggested oral diseases (periodontitis (PD) and dental caries) may increase susceptibility to bone loss. However, inherent confounding of observational studies limits causal interpretation. We aimed to conduct Mendelian randomization (MR) analysis to estimate the causal effect of oral diseases on osteoporosis (OP), bone mineral density (BMD) and fracture risk. Methods We used summary-level GWAS meta-analysis data from the GLIDE consortium to identify 7 and 17 single-nucleotide polymorphisms (SNPs) for periodontitis and DMFS (the sum of Decayed, Missing, and Filled tooth Surfaces) as the instrumental variables. MR analyses of these instruments were performed on European individuals for the association with BMD of forearm, femoral neck and lumbar spine; and individuals from FinnGen consortium for OP, OP with pathological fracture, postmenopausal OP with pathological fracture, and site-specific fractures. We performed single-variable Mendelian randomization (SVMR) and multivariable Mendelian randomization (MVMR) to simultaneously assess independent causal effects of PD and DMFS on different outcomes. The estimates were primarily derived using inverse variance weighted (IVW) methods. Sensitivity analyses included weighted median, MR-egger, and Leave-one-out test. Results In MVMR, after adjusting for PD, DMFS has a positive causal effect osteoporosis (OR = 1.165, [95% CI 1.020 to 1.331, P = 0.025]) and postmenopausal OP with pathological fracture (OR = 1.422, [95% CI 1.027 to 1.969, P = 0.034]). However, these causal relationships were not observed in the single-variable Mendelian randomization (SVMR) analysis. The causal associations were robust in various sensitivity analyses. Conclusions In conclusion, dental caries causally increases the risk of OP and postmenopausal OP with pathological fracture, suggesting the existence of teeth-bone axis. Proactive osteoporosis screening in patients with severe dental caries may be warranted for clinical consideration.