Background <p>Whether common sleep phenotypes causally influence anorexia nervosa (AN) or obsessive–compulsive disorder (OCD) is unclear; we used Mendelian randomization (MR) to test effects of daytime sleepiness (DS), insomnia, and sleep apnea (SA) on AN and OCD, and the reverse direction.</p> Methods <p>We conducted bidirectional two-sample MR using genome-wide association studies (GWAS) of DS (<i>N</i> = 452,071), insomnia (<i>N</i> = 453,379), SA (<i>N</i> = 523,366), AN (<i>N</i> = 72,517), and OCD (<i>N</i> = 33,943), predominantly of European ancestry. Instruments were single-nucleotide polymorphisms meeting genome-wide significance (<i>p</i> &lt; 5 × 10^⁻8), clumped for linkage disequilibrium (<i>r</i>² &lt; 0.001) with mean <i>F</i> &gt; 10; harmonization aligned effect alleles. The primary estimator was inverse-variance weighted (IVW), with weighted median and MR-Egger as sensitivity analyses. Heterogeneity and pleiotropy were evaluated with Cochran’s Q, MR-Egger intercepts, and MR-PRESSO.</p> Results <p>Genetic liability to SA was associated with higher risks of AN (IVW OR = 1.47, 95% CI 1.13–1.91, <i>p</i> = 0.004) and OCD (IVW OR = 1.22, 95% CI 1.08–1.37, <i>p</i> = 0.001). DS and insomnia showed no clear associations with either outcome. Reverse analyses provided no evidence that genetic liability to AN or OCD influenced DS, insomnia, or SA. Sensitivity analyses indicated adequate instrument strength (mean <i>F</i> = 15.55–25.75), limited directional pleiotropy, and estimates that were directionally consistent across methods; MR-PRESSO signals in models with AN did not materially alter conclusions.</p> Conclusions <p>Genetic evidence supports a causal contribution of SA, not DS or insomnia, to AN and OCD, with no evidence for reverse causation. Sleep-disordered breathing may be a modifiable target for prevention or adjunctive treatment; replication in diverse ancestries using objective sleep measures is warranted.</p> <p>Level of evidence III well-designed cohort analytic study.</p>