Apolipoprotein-E epsilon 4 (APOE-epsilon 4) has been consistently associated with Alzheimer disease (AD) and may be responsible for an earlier age at onset. We have previously reported a diminished association between APOE-epsilon 4 and AD in African Americans. Using a new method, which allows inclusion of censored information, we compared relative risks by APOE genotypes in an expanded collection of cases and controls from three ethnic groups in a New York community. The relative risk for AD associated with APOE-epsilon 4 homozygosity was increased in all ethnic groups (African American relative risk [RR]=3.0; 95% confidence interval [CI]=1.5-5.9; Caucasian RR=7.3, 95% CI=2.5-21.6; and Hispanic RR=2.5, 95% CI=1.1-5.7), compared with those with APOE-epsilon 3/epsilon 3 genotypes. The risk was also increased for APOE-epsilon 4 heterozygous Caucasians (RR=2.9, 95% CI=1.7-5.1) and Hispanics (RR=1.6, 95% CI=1.1-2.3), but not for African Americans (RR=0.6, 95% Ci=0.4-0.9). The age distribution of the proportion of Caucasians and Hispanics without AD was consistently lower for APOE-epsilon 4 homozygous and APOE-epsilon 4 heterozygous individuals than for those with other APOE genotypes. In African Americans this relationship was observed only in APOE-epsilon 4 homozygotes. These results confirm that APOE genotypes influence the RR of AD in Caucasians and Hispanics. Differences in risk among APOE-epsilon 4 heterozygote African Americans suggest that other genetic or environmental factors may modify the effect of APOE-epsilon 4 in some populations.