In recent times, the association between HF and BMD has attracted enormous interest in the scientific community. However, published epidemiological observational studies on the relationship between HF and BMD remain inconclusive. Herein, we evaluated from the analytical perspective a two-sample bidirectional MR study to analyze the causal association between HF and BMD using a summary-level GWAS Catalog. To select instrumental SNPs strongly associated with exposure, we took a series of rigorous quality control steps at the time of analysis. The causal MR assessment of HF on the risk of BMD was performed first and then in the opposite direction. To make the conclusions more reliable and robust, the fixed-effects IVW, weighted median-based method, MR–Egger, simple mode and weighted mode were utilized. A maximum likelihood model was also used if necessary. MR–Egger regression, IVW “leave-one-out” sensitivity analysis, MR-PRESSO, MR–Egger intercept test and Cochran's Q statistic methods were used to assess heterogeneity and pleiotropy. Our MR studies supported a meaningful causal association between HF and TB-BMD (IVW: OR = 0.78, 95% CI: 0.68–0.87, p = 0.00588). At the same time, we did not find a significant causal relationship between HF and FA-BMD, FN-BMD or LS-BMD. No significant causal relationships between BMD and HF were observed. This bidirectional MR analysis suggested a causal association of HF with only low TB-BMD, while the reverse causality hypothesis was not found. Studies of the prevention and treatment of total bone mineral density decline in patients with heart failure need to be performed.