Background: Apolipoprotein E (APOE) is the most statistically significant genetic risk factor for late-onset Alzheimer disease (LOAD). The linkage disequilibrium pattern around the APOE gene has made it difficult to determine whether all the association signal is derived from APOE or whether there is an independent signal from a nearby gene. Objective: To attempt to replicate a recently reported association of APOE 3-TOMM40 haplotypes with risk and age at onset. Design: We used standard techniques to genotype several polymorphisms in the APOE-TOMM40 region in a large case-control series, in a series with cerebrospinal fluid biomarker data, and in brain tissue. Setting: Alzheimer's Disease Research Center. Participants: Research volunteers who were cognitively normal or had Alzheimer disease. Main Outcome Measures: Disease status and age at onset. Results: We did not replicate the previously reported association of the polyT polymorphism (rs10524523) with risk and age at onset. We found a significant association between rs10524523 and risk of LOAD in APOE 33 homozygotes but in the opposite direction as the previously reported association (the very long allele was underrepresented in cases vs controls in this study (P=.004]). We found no association between rs10524523 and cerebrospinal fluid tau or β-amyloid 42 levels or TOMM40 or APOE gene expression. Conclusions: Although we did not replicate the earlier association between the APOE 3-TOMM40 haplotypes and age at onset, we observed that the polyT polymorphism is associated with risk of LOAD in APOE 33 homozygotes in a large case-control series but in the opposite direction as in the previous study.