Studies of the effect of alcohol consumption on cognitive ability are often confounded. One approach to avoid confounding is the Mendelian randomization design. Here, we used such a design to test the hypothesis that a genetic score for alcohol processing capacity moderates the association between alcohol consumption and lifetime change in cognitive ability. Members of the Lothian Birth Cohort 1936 completed the same test of intelligence at age 11 and 70 years. They were assessed for recent alcohol consumption in later life and genotyped for a set of four single-nucleotide polymorphisms in three alcohol dehydrogenase genes. These variants were unrelated to late-life cognition or to socioeconomic status. We found a significant gene × alcohol consumption interaction on lifetime cognitive change (p = 0.007). Individuals with higher genetic ability to process alcohol showed relative improvements in cognitive ability with more consumption, whereas those with low processing capacity showed a negative relationship between cognitive change and alcohol consumption with more consumption. The effect of alcohol consumption on cognitive change may thus depend on genetic differences in the ability to metabolize alcohol.