OBJECTIVES: To examine the relationship between alcohol consumption and cognitive function in the middle-aged and older population through the HRS data. METHODS: A cross-sectional study was conducted using 2012 HRS telephone survey data. Survey respondents who finished all relevant questions in the cognition section and provided information on alcohol consumption were included in our analysis. Cognition scores (0-35) were summed for each individual, indicating overall cognitive performance. A multivariate regression model, including age, years of education, and general health status was used to estimate the effects of drinking status (abstainer/drinker), drinking days/week, and binge drinking (>4 drinks per occasion) on cognitive performance. The model was stratified by gender. Curvilinear relationships were also examined because previous studies demonstrated nonlinear relationships between drinking and cognition. RESULTS: There were 7625 participants included (male, 42.8%; female, 57.2%). Overall, females scored 0.97 higher than males (SE=0.09, p<0.001). For males, we identified 1303 (39.3%) abstainers and the mean drinking days/week was 1.5 days for drinkers; for females, we identified 2405 (55.1%) abstainers and the mean drinking days/week was 2.0 days for drinkers. No significant curvilinear relationship was observed for cognitive function and drinking days/week among either males or females. Among males, abstainers scored 0.91 (SE=0.18, p<0.001) lower compared with non-binge drinkers. Among male drinkers, one more drinking day/week was associated with a significant increase of 0.15 (SE=0.05, p<0.01) points. Among females, abstainers scored 0.95 (SE=0.17, p<0.001) lower than non-binge drinkers, but among drinkers, drinking days/week was not associated with cognition. Binge drinking (4.1% of males, 0.5% of females) was not significantly related to cognition.  CONCLUSIONS: Our findings suggest a positive effect of drinking on cognition, however this effect is very small and varied by gender. Longitudinal studies are needed to investigate the cumulative effect of alcohol consumption on cognition and to specify the causal relationship.