BackgroundObservational studies have found adiponectin inversely associated with cardiovascular disease thereby indicating a potential target of intervention, but genetically higher adiponectin appears unrelated to coronary artery disease (CAD). To clarify, we examined the role of genetically predicted adiponectin in CAD in a larger study and additionally examined the role of genetically predicted CAD in adiponectin using a bi-directional Mendelian randomization study. MethodsWe obtained estimates using inverse variance weighting (IVW) with multiplicative random effects, based on 21 single nucleotide polymorphism (SNPs) to predict adiponectin and 28 SNPs for CAD, using two large genome wide association studies of adiponectin (ADIPOGen Consortium (n = 39,883)) and CAD CARDIoGRAMplusC4D 1000 Genomes based GWAS (n = 60,801 CAD cases; n = 123,504 controls). Sensitivity analyses included using MR-Egger, weighted median method, and exclusion of potentially invalid (pleiotropic) SNPs. ResultsAdiponectin was inversely associated with CAD (odds ratio 0.82 per unit increase log transformed adiponectin, 95% confidence interval (CI) 0.71 to 0.94) using IVW. However, the association was null using a weighted median method or MR-Egger, or after exclusion of pleiotropic SNPs acting on obesity related traits. CAD was not associated with adiponectin (−0.011 log transformed adiponectin unit per log odds CAD, 95% CI −0.039 to 0.017), with similar findings from MR-Egger, weighted median method or exclusion of pleiotropic SNPs. ConclusionAdiponectin is unlikely a cause of CAD although we cannot completely rule out the possibility. Previous observational studies are likely driven by factors driving both adiponectin and CAD, whose elucidation might provide new insights concerning interventions for CAD.