Many observation studies have demonstrated a close relationship between rheumatoid arthritis (RA) and osteoporosis (OP). However, the causal genetic correlation between RA and OP remains unclear. In this study, we performed bi-directional Mendelian randomization (MR) analyses to explore causal inference between these two traits. The instrumental variables for RA were selected from a large-scale genome-wide association study (GWAS) (1,523 cases and 461,487 controls). Bone mineral density (BMD) at five different sites (heel (n=265,627), forearm (FA) (n=8,143), femoral neck (FN) (n=32,735), lumbar spine (LS) (n=28,498), and total body (n=28,498)) were used as phenotypes for OP. The inverse variance weighted (IVW) method did not detect any causal effect of BMDs on RA except heel BMD (beta = -7.57 × 10-4, p = 0.02). However, other methods (MR-Egger, weighted median, weighted mode, MR-PRESSO, and MR-RAPS) showed no causal association between heel BMD and RA. Likewise, we did not find a causal effect of RA on BMD at any sites. In conclusion, we found no evidence that RA is causally associated with OP/BMD, or vice versa. We suggested that the associations found in previous observational studies between RA and OP/BMD are possibly related to secondary effects such as antirheumatic treatment and reduced physical activity.