ObjectiveTo explore the causal association between venous thromboembolism and cardiovascular disease risks using a two-sample bidirectional Mendelian randomization study. MethodsThe single-nucleotide polymorphism (SNP) data associated with venous thromboembolism (VTE) and cardiovascular diseases (CVD) from genome-wide association studies were obtained as instrumental variables. Inverse variance weighted (IVW) was used as the main Mendelian randomization method and other methods were used as supplementary methods. Cochran's Q test, the intercept term of MR-Egger, and MR-PRESSO were used to assess pleiotropy and heterogeneity to ensure the robustness of the results. ResultsThe IVW method suggested a causal association between venous thromboembolism and atrial fibrillation (OR=1.033, 95%CI 1.009 to 1.058, P=0.008), but no association was identified between VTE and coronary artery disease (OR=0.994, 95%CI 0.974 to 1.023, P = 0.551), heart failure (OR=1.021, 95%CI 0.992 to 1.050, P=0.159) and myocardial infarction (OR=1.012, 95%CI 0.971 to 1.055, P=0.568). The results of Cochran's Q test showed that there was no heterogeneity in the MR analyses of VTE and CVD. The MR-Egger intercept analysis and the MR-PRESSO global testing did not detect potential horizontal pleiotropy, and the results were robust. Reverse Mendelian randomization analysis was used to verify the presence of reverse causal associations. The reverse MR analysis demonstrated that reverse causal associations between venous thromboembolism and cardiovascular diseases were not evidenced. ConclusionThe results of the Mendelian randomization study demonstrated a causal association between venous thromboembolism and atrial fibrillation, but not with CAD, HF, or MI, providing a new perspective on the prevention and treatment of atrial fibrillation.