BACKGROUND: Observational studies have demonstrated associations between autoimmune thyroid disease and both mood disorders and dementia, but the direction of causality has not been established. METHODS: We employed bidirectional, two-sample Mendelian randomization analyses to assess the causal links between autoimmune thyroid disease (Graves' disease and autoimmune thyroiditis), and mood disorders (major depressive disorder and bipolar disorder) as well as dementia (Alzheimer's disease and dementia with Lewy bodies). We obtained Genome-Wide Association Study data from the Finngen database and IEU Open GWAS. For statistical analysis, we utilized several robust methods inverse variance weighted, MR-Egger, weighted median, simple mode and weighted mode. Sensitivity analyses are performed to detect potential horizontal pleiotropy, and to exclude any outlier data points that could skew our results. RESULTS: Gene-proxy for Graves' disease was associated with an increased risk of major depressive disorder (P-value = 0.02), whereas gene-proxy for autoimmune thyroiditis was linked to a greater risk of bipolar disorder (P-value = 0.043). The levels of pleiotropy and heterogeneity among genetic variants suggest bias is unlikely, leave-one-out test confirmed a robust correlation. No indication that mood disorders influence the risk of autoimmune thyroid disease, and no evidence of a causal relationship between autoimmune thyroid disease and dementia. CONCLUSIONS: Graves' disease is associated with heightened risk of major depressive disorder, and autoimmune thyroiditis is correlated with increased risk of bipolar disorder.