OBJECTIVE: To examine the association of plasma cholesterol levels, lipid-lowering agent (LLA) intake, and APOE genotype with dementia prevalence. METHODS: The Three-City Study is a population-based cohort of 9,294 subjects selected from the electoral rolls of three French cities (Bordeaux, Dijon, Montpellier). Baseline examination included extensive assessment of exposure to vascular risk factors (including cholesterol levels and LLA use [statin or fibrate]) and clinical diagnosis of dementia. RESULTS: Two percent of participants were demented at baseline. Overall 32.4% of participants had hyperlipidemia, and 15.6% were prescribed statins and 13.7% fibrates. After adjusting for age, gender, education level, and study center, the odds ratio (OR) for dementia was observed to be lower among LLA users (OR = 0.61, 95% CI = 0.41 to 0.91) compared with subjects taking no LLAs. There was no differential effect between statin and fibrate users. The odds for dementia were increased in subjects with hyperlipidemia (OR = 1.43, 95% CI = 1.03 to 1.99). Further adjustment for potential confounders did not modify these associations. In addition, the association between LLA intake and dementia was not modified by APOE genotype, whereas hyperlipidemia was significantly associated with increased dementia prevalence only in non-epsilon4 carriers and non-Alzheimer disease cases. Finally, in participants taking LLAs, the odds for dementia were decreased only in those having normal lipid levels. CONCLUSIONS: This observational study provides further evidence that lipid-lowering agents are associated with decreased risk of dementia, whereas hyperlipidemia is associated with increased odds for non-Alzheimer-disease-type dementia. These effects appear to be independent of all major potential confounders.