There is little information on the association of the APOEe4 allele and AD risk in African populations. In previous analyses from the Indianapolis-Ibadan dementia project, we have reported that APOE ε4 increased the risk for Alzheimer's disease (AD) in African Americans but not in Yoruba. This study represents a replication of this earlier work using enriched cohorts and extending the analysis to include cognitive decline. In this longitudinal study of two community dwelling cohorts of elderly Yoruba and African Americans, APOE genotyping was conducted from blood samples taken on or before 2001 (1,871 African Americans & 2,200 Yoruba). Mean follow up time was 8.5 years for African Americans and 8.8 years for Yoruba. The effects of heterozygosity or homozygosity of ε4 and of the possession of e4 on time to incident AD and on cognitive decline were determined using Cox's proportional hazards regression and mixed effects models. After adjusting for covariates, one or two copies of the APOE ε4 allele were significant risk factors for incident AD (p < 0.0001) and cognitive decline in the African-American population (p < 0001). In the Yoruba, only homozygosity for APOE ε4 was a significant risk factor for AD (p = 0.0002) but not for cognitive decline (p = 0.2346), however, possession of an e4 allele was significant for both incident AD (p = 0.0489) and cognitive decline (p = 0.0425). In this large longitudinal comparative study, APOE ε4 had a significant, but weaker, effect on incident AD and on cognitive decline in Yoruba than in African Americans. The reasons for these differences remain unclear.