Project Description: A J-shaped relationship between mean alcohol consumption and risk of cognitive decline has been noted repeatedly but we know little about how other potential harmful drinking patterns, such as getting drunk, affect cognitive function over time. We used data on adults aged ≥65 from the English Longitudinal Study of Ageing (ELSA), a population-based study of community-dwelling older adults in England. Between 1998 and 2001 participants were asked how many times they had been drunk in the preceding three months. A battery of cognitive function tests was administered in 2002 and again in 2008. We examined the relationship between frequency of drunkenness and change in cognitive function scores between 2002 and 2008 (using linear regression models) and being in the lowest 25% of cognitive function scores in 2008 (using generalized linear models to calculate relative risks). Models were adjusted for potential confounding by age, gender, level of education, smoking status, number of depressive symptoms, and mean baseline level of alcohol consumption. Among 1,221 drinkers, 48 (3.9%) reported having been drunk one or more times in the preceding three months. In adjusted models, compared to those who had had not been drunk in the preceding three months those who had been drunk two or three times had a follow-up cognitive function score -0.23 standard deviations lower (95% Confidence Interval [CI] -0.69 to 0.23) and those who had been drunk four or more times had a score 0.51 standard deviations lower (95% CI -0.95 to -0.08). The relative risk of being in the lowest quarter of scores at follow-up associated with being drunk two or three times was 2.12 (95% CI 0.62 to 7.27) and 3.35 (95% CI 1.15 to 9.80) with being drunk four or more times. These results suggest frequent episodes of drunkenness in older adults are associated with a greater risk of cognitive decline. Given the likelihood of underreporting and underrepresentation biases in our data these findings warrant attention as well as replication in other cohorts.