In order to elucidate the correlation and its pathways between liver diseases and vascular dementia (VaD). We performed two-sample Mendelian randomization (MR) study. We performed a two-sample MR study using summary-level genome-wide association study (GWAS) data for 4 liver diseases (nonalcoholic fatty liver disease [NAFLD], nonalcoholic steatohepatitis [NASH], alcoholic liver disease [ALD], and fibrosis/cirrhosis [FACOL]) and VaD. Inverse variance weighted (IVW) served as the primary analysis, complemented by a suite of sensitivity analyses (Cochran Q test, MR-Egger regression and MR pleiotropy residual sum and outlier [MR-PRESSO]). Furthermore, multivariable MR (MVMR) was conducted to adjust for key metabolic traits, including fasting glucose and triglycerides. The IVW analysis indicated that no significant causal associations were found between any of the 4 liver diseases and the risk of VaD. Sensitivity analyses indicated that the results for NAFLD and FACOL may be affected by heterogeneity, and NAFLD may also be affected by horizontal pleiotropy. Crucially, after adjusting for genetic liability to elevated triglycerides or fasting glucose in MVMR analyses, the initially null association for NAFLD became significant, revealing a protective effect against VaD. Associations for other liver diseases remained null after adjustment. This MR study found no direct causal link between liver diseases and VaD. However, adjusting for glucose or triglycerides revealed a potential protective effect of NAFLD, suggesting metabolic factors may confound the association. Further research is warranted to disentangle the complex relationship between these health conditions.