In a cohort study, 1,301 subjects free of dementia at baseline in the Kungsholmen Project were followed up to 6 years. We studied the association between use of aspirin and non-steroidal anti-inflammatory drugs (NSAIDs), incidence of Alzheimer's disease (AD) and overall dementia, and the influence of the apolipoprotein E epsilon4 allele. In stratified analyses, a relative risk (RR) of 1.80 (95% CI 1.14-2.83) for AD was seen, in the apoE epsilon4-negative group using aspirin. This implicates a possible different mechanism of developing AD in this group. We also found a possible protective effect of NSAIDs against AD, since no one who used NSAIDs for around 3 years had developed AD 3 years later. One user developed vascular dementia, and a low point value of risk was seen, however, not significant (RR 0.23; 95% CI 0.03-1.68). This could be due to the small samples in our study, or to comorbidity contributing to the development of dementia in this elderly population.