We determined whether higher endogenous estradiol levels were associated with lower risk of dementia in older men and women not using hormonal replacement therapy, using a case-cohort design within the Rotterdam Study, a population-based follow-up study on chronic diseases, including dementia, in 7983 subjects aged 55 years or older, and ongoing since 1990. The analyses were based on a random subcohort of 508 women and 438 men, and on 76 women and 53 men with incident dementia. Cox proportional hazards models with robustly estimated standard errors showed that in women higher levels of total estradiol were associated with higher risk of dementia (age-adjusted hazard ratio per standard deviation increase 1.38; 95% CI 1.04-1.84). Age-adjusted HR's of Alzheimer's disease and vascular dementia associated with higher levels of total estradiol (per SD increase) were 1.24 (95% CI 0.87-1.76) and 2.19 (95% CI 1.22-3.92), respectively. Similar results were observed for bioavailable estradiol. Additional adjustments for potential confounders did not change the results substantially. In men, no clear association was observed between estradiol levels and risk of dementia or its subtypes. The findings do not support the hypothesis that higher levels of endogenous estradiol reduce risk of dementia, neither in women nor in men.