AIMS: To use Mendelian randomization to assess whether alcohol intake was causally associated with cognitive function. DESIGN: Mendelian randomization using a genetic variant related to alcohol intake (ADH1B rs1229984) was used to obtain unbiased estimates of the association between alcohol intake and cognitive performance. SETTING: Europe. PARTICIPANTS: More than 34 000 adults. MEASUREMENTS: Any versus no alcohol intake and units of intake in the previous week was measured by questionnaire. Cognitive function was assessed in terms of immediate and delayed word recall, verbal fluency and processing speed. FINDINGS: Having consumed any versus no alcohol was associated with higher scores by 0.17 standard deviations (SD) [95% confidence interval (CI) = 0.15, 0.20] for immediate recall, 0.17 SD (95% CI = 0.14, 0.19) for delayed recall, 0.17 SD (95% CI = 0.14, 0.19) for verbal fluency and 0.12 SD (95% CI = 0.09, 0.15) for processing speed. The minor allele of rs1229984 was associated with reduced odds of consuming any alcohol (odds ratio = 0.87; 95% CI = 0.80, 0.95; P = 0.001; R(2)  = 0.1%; F-statistic = 47). In Mendelian randomization analysis, the minor allele was not associated with any cognitive test score, and instrumental variable analysis suggested no causal association between alcohol consumption and cognition: -0.74 SD (95% CI = -1.88, 0.41) for immediate recall, -1.09 SD (95% CI = -2.38, 0.21) for delayed recall, -0.63 SD (95% CI = -1.78, 0.53) for verbal fluency and -0.16 SD (95% CI = -1.29, 0.97) for processing speed. CONCLUSIONS: The Mendelian randomization analysis did not provide strong evidence of a causal association between alcohol consumption and cognitive ability.