To verify whether chronic use of nonsteroidal anti-inflammatory drugs (NSAIDs) has a protective effect against cognitive decline in older persons. Prospective study with a 3-year observation period. Three communities of the Established Populations for Epidemiologic Studies of the Elderly (EPESE). A population-based sample of 7671 subjects who received an in-person interview at the sixth annual follow-up. Persons with documented NSAID use at the time of the interview and 3 years before were considered chronic users (21%), while all other persons were considered as nonusers. Change over time in cognitive function assessed as the number of correct answers to a 9-item version of the Short Portable Mental Status Questionnaire (SPMSQ). For every level of SPMSQ score measured at the beginning of the observation period, the mean SPMSQ score after 3 years was higher in chronic NSAID users than in nonusers. Cognitive function at the end of the observation period was significantly higher in chronic NSAID users than in controls, adjusting for initial SPMSQ score and potential confounders. Older age, female gender, education, and history of cerebrovascular disease were also independent predictors of lower SPMSQ score. In the multivariate analysis, the magnitude of the protective effect estimated for NSAID use was comparable to a difference in age of 3.5 years. The percentage of persons who started above a specific SPMSQ score cut-point and deteriorated below that cut-point over a 3-year period was significantly lower in chronic NSAID users than in nonusers (30.2% vs 34.3%, P = .03, for decline below SPMSQ score of 8 and 12.3% vs 14.4% for decline below SPMSQ score of 6, P = .04). After controlling for potential confounders, the relative risk of cognitive declining in chronic NSAID users compared with nonusers was 0.82 (95% Confidence Interval: 0.69-0.98) for a decline below a score of 8, and 0.80 (95% CI: 0.66-0.98) for a decline below a score of 6. These results support the association between NSAID use and reduction in cognitive decline in older persons. Ultimately, randomized controlled trials must be done to prove a beneficial effect definitively.