Background Growing evidence suggests that atopic dermatitis (AD) is associated with an increased risk of depressive disorders and anxiety. However, existing studies were observational and may have uncovered correlations but could not easily disentangle noncausal or reverse-causal associations because these associations could be confounded and may not reflect true causal relationships. Objectives To examine, in a two-sample Mendelian randomization study, the potential effect of AD on the risk of depressive disorders and anxiety. Methods Genetic instruments from the largest available genome-wide association study (GWAS) for AD (10 788 cases and 30 047 controls) were used to investigate the relationship to broad depression (170 756 cases and 329 443 controls), major depressive disorder (MDD; 30 603 cases and 143 916 controls) and anxiety (5580 cases and 11 730 controls). A set of complementary approaches were carried out to assess horizontal pleiotropy and related potential caveats occurring in MR studies. Results We observed no causal impact of AD on the risk of depressive disorders and anxiety, with close-to-zero effect estimates. The inverse weighted method revealed no associations of AD on broad depression [odds ratio (OR) 1 center dot 014; P = 0 center dot 431], probable MDD (OR 1 center dot 002; P = 0 center dot 568), International Classification of Diseases, Ninth/Tenth Revision-based MDD (OR 1 center dot 001; P = 0 center dot 466) or anxiety (OR 1 center dot 097; P = 0 center dot 180). Conclusions This MR study does not support a causal effect of AD on depression and anxiety.