CONTEXT: Atrial fibrillation (AF), cardiac arrhythmias, and related risk factors are common in patients with Cushing's syndrome, or clinical chronic hypercortisolism. While hypercortisolism may be associated with AF, this association has not yet been ascertained causally. OBJECTIVE: To determine whether plasma cortisol is causally associated with AF using a 2-sample Mendelian randomization (MR) design. METHODS: Three genetic variants in the SERPINA1/SERPINA6 locus and functionally associated with plasma cortisol were identified in the CORtisol NETwork consortium (12 597 participants). Summary-level genome-wide association study (GWAS) data for the associations between the cortisol-associated variants and AF were obtained from a GWAS meta-analysis of 6 studies (60 620 AF cases and 970 216 noncases) and the FinnGen consortium (17 325 AF cases and 97 214 noncases). The fixed-effects inverse-variance weighted approach accounting for genetic correlations between variants was used for analysis. Multivariable MR analyses were conducted to assess potential mediating effects of systolic blood pressure (SBP) and waist circumference (WC). Summary-level GWAS data for SBP and WC were obtained respectively from the International Consortium of Blood Pressure (757 601 participants) and the Genetic Investigation of ANthropometric Traits consortium (232 101 participants). RESULTS: One standard deviation increase in genetically predicted plasma cortisol was associated with greater risk of AF (odds ratio [OR] 1.20, 95% CI 1.06-1.35). The association attenuated when adjusting for genetically predicted SBP and WC (OR 0.99, 95% CI 0.72-1.38). CONCLUSION: Evidence derived from the MR study suggests a positive association between plasma cortisol and risk of AF, likely mediated through SBP and WC.