Carriers of the apolipoprotein E (APOE) epsilon4 allele show significantly higher risk of Alzheimer disease (AD). The aim of this present study was to test the hypothesis that a significant interaction exists between APOE genotype and gender on AD. Interactions of epsilon4 by gender, although indicated in the literature, require further verification. A total of 195 past or current control or AD participants in an ongoing longitudinal study of aging and dementia were genotyped. All subjects were at least 60 years old; demented subjects met clinical or pathologic criteria for late-onset AD. Logistic regression analysis and proportional hazard models were used to evaluate joint effects of APOE and gender. A significant statistical interaction between APOE and gender was shown (p = 0.04) in logistic regression analysis. Women carrying one or more APOE-epsilon4 allele were more likely to develop AD [odds ratio (OR) = 7.8, 95% confidence interval (CI) = 3.2-19. 1]. For men, the presence of the APOE-epsilon4 allele was not associated with a statistically significant increased risk (OR = 1.6, 95% CI = 0.5-5.3). The interaction term in the proportional hazards model neared (p = 0.07) statistical significance, and a similar but reduced gender effect was shown. The analysis suggests that the presence of one or more APOE-epsilon4 allele confers a substantially greater risk of AD to women than to men. These findings in part may account for reports of increased risk of AD faced by women.