Objective: To investigate the protective effect of NSAIDs and aspirin separately on cognitive decline in elderly subjects, controlling for consistent use of these agents over a prolonged period of time. Methods: The study sample consisted of 1007 subjects, drawn from a population-based random sample of elderly individuals, 62-85 years old, who participated in a 3-year follow-up study. From this sample subjects were selected, who did use NSAIDs and completed all cognitive tests at both measurements (n=137), and subjects who did not use NSAIDs and completed all cognitive tests (n=475). Cognitive tests included the Mini-Mental State Examination (MMSE), tests for episodic memory (Auditory Verbal Learning Test) and information processing speed (coding task). Cognitive decline was computed using Edwards-Nunnally method. Multiple logistic regression analyses were performed to examine the association between NSAID (with and without aspirin) and decline in cognitive performance. Besides, the interaction of NSAIDs with age on cognitive decline was determined. Results: The relative risk estimates of decline in episodic memory (immediate recall) adjusted for age, gender, education, baseline MMSE, vascular diseases, diabetes mellitus and (rheumatoid) arthritis for aspirin users only was more than three times reduced (OR: 0.30, 95% CI: 0.09-0.82). The odds ratio for decline in memory of NSAID use without aspirin, adjusted for age, gender, education, baseline MMSE, vascular diseases, diabetes mellitus and (rheumatoid) arthritis was not significant (OR: 1.00, 95% CI: 0.39-2.93). The effect of aspirin was significant only in persons of 75 years and over (OR: 0.10, 95% CI: 0.01-0.81), not in subjects younger than 75 years (OR: 0.52, 95% CI: 0.14-1.96). NSAIDs did not have benefit on information processing speed. In 92% of aspirin users a low dose of 100mg daily or less was used. Conclusion: Low-dose aspirin might be protective for decline in memory in individuals of 75 years and over. The benefit of a low-dose aspirin does not support an anti-inflammatory effect, but suggests an antiplatelet effect. Therefore, a possible protective effect of low-dose aspirin on cognitive decline is likely only in subjects with aspirin use over a prolonged period of time. © 2002 Elsevier Science Inc. All rights reserved.