Alzheimer disease (AD) is an emotionally devastating and exceptionally costly disease. Apolipoprotein E (APOE) is a major risk factor gene for AD regardless of age of onset or family history. However, this association may not be as strong or consistent in ethnic groups such as African Americans, raising the possibility of other modifier gene(s). In a group of African American AD patients, a significantly increased risk of AD was associated with two E4 alleles (OR = 5.6; 95% CI = 1.5–21.0) or one E4 allele (OR = 2.5; 95% CI = 1.3–5.0) when compared to E3/E3 genotype, and there was a significant lowering of age of onset for affecteds with E4/E4 genotype as compared to one E2 allele (P = 0.02) or all others (P = 0.03). We also found a significant increase in age of onset with the −308 #2 (A) allele of TNF when compared to AD cases with no #2 allele. A significant increase in age was also demonstrated with the #2 allele (99 base pairs) of the microsatellite TNFa, located ∼ 10.5 kb upstream of TNF. When these two alleles were combined with the TNF −238G (#1) allele to give a haplotype, the significant increase in age was still demonstrated. Polymorphisms in the APOE promoter and six other candidate genes did not appear to demonstrate any significant association with our African American AD patients. Our results confirm the established association of APOE4 to AD observed in several ethnic groups, including African Americans. In addition, TNF appears to have some modifying effect in AD, primarily on age of onset, or it could be in linkage disequilibrium with a modifier locus nearby. © 2001 Wiley-Liss, Inc.