Numerous studies have shown that the risk of Alzheimer's disease (AD) is associated with the dose of the epsilon 4 allele of apolipoprotein E (ApoE). However, more than one third of AD patients lack epsilon 4 and many persons having epsilon 4 survive cognitively intact to old age. We evaluated the lifetime risk of disease in 3,999 first-degree relatives of 549 probands who met the criteria for probable or definite AD and whose ApoE genotypes were known. ApoE genotypes for relatives were not determined. After age 65 the risk among relatives was proportional, as much as 7 to 10% at age 85, to the number of epsilon 4 alleles present in the proband. Risks to relatives of ApoE 2/2 and 2/3 probands were nearly identical at all ages to risks for relatives of ApoE 3/3 probands. The expected proportion of relatives having at least one epsilon 4 allele was calculated for each genotype group based on the distribution of parents, sibs, and offspring in the sample. Among relatives in the ApoE 3/3 group, the lifetime risk for AD by age 90 was three times greater than the expected proportion of epsilon 4 carriers, suggesting that factors other than ApoE contribute to AD susceptibility. Furthermore, the 44% risk of AD by age 93 among relatives of ApoE 4/4 probands indicates that as many as 50% of people having at least one epsilon 4 allele do not develop AD. We also found that among male relatives, risk of AD in the ApoE 3/4 group was similar to that for the ApoE 3/3 group but significantly less than the risk for the ApoE 4/4 group. In contrast, among female relatives the risk for the ApoE 3/4 group was nearly twice that for the ApoE 3/3 group and identical to the risk for the ApoE 4/4 group. These findings are consistent with a sex-modification effect of the E4 isoform on disease susceptibility.