Objective: To examine the prevalence of Alzheimer's disease (AD) and other dementias in relation to age, education, sex, and genotype at APOE. Recent studies suggest age heterogeneity in the risk of AD associated with the APOE genotype and a possible interaction between APOE-ε4 and female sex as risk factors. We studied these topics in the 5,677 elderly residents of Cache County, Utah, a population known for long life expectancy and high participation rates. Methods: We screened for dementia with a brief cognitive test and structured telephone Dementia Questionnaire, then examined all individuals with apparent cognitive symptoms and a sample of others. We estimated age-specific prevalence of AD and other dementias and used multiple logistic regression models to describe relation of AD prevalence to age, sex, education, and APOE genotype. Results: We found 335 demented individuals, 230 (69%) with definite, probable, or possible AD (positive predictive value versus autopsy confirmation 85%). The adjusted prevalence estimate for AD was 6.5% and for all dementias 9.6%. After age 90, the adjusted prevalence estimate for AD was 28% and for all dementias 38%. Regression models showed strong variation in AD prevalence with age, sex, education, and number of ε4 alleles (effect of ε2 not significant). Models were improved by a term for age-squared (negative coefficient) and by separate terms for interaction of age with presence of one or two ε4 alleles. An association of AD with female sex was ascribable entirely to individuals with ε4. Conclusions: In participants with no ε4 alleles, the age-specific prevalence of AD reached a maximum and then declined after age 95. In ε4 heterozygotes a similar maximum was noted earlier at age 87, in homozygotes at age 73. Female sex was a risk factor for AD only in those with ε4. The ε4 allele accounted for 70% of the population attributable risk for AD.