Recent studies have found an association between presence of apolipoprotein E (APOE) epsilon 4 allele and Alzheimer's disease (AD). The present study compared the cumulative risk of primary progressive dementia (PPD) in relatives of AD probands carrying at least one copy of the epsilon 4 allele with the relatives of AD probands not carrying epsilon 4 and with relatives of non-demented controls. Our aim was to determine whether the familial aggregation of PPD in relatives of AD probands is primarily due to those carrying epsilon 4. Seventy-seven neuropathologically diagnosed AD patients were obtained as probands through our Alzheimer's Disease Research Center Brain Bank. AD probands were genotyped for APOE. As a comparison group, 198 non-demented probands were also included. Through family informants, demographic and diagnostic data were collected on 382 first-degree relatives (age > or = 45 years) of AD probands and 848 relatives of the controls. We found that the cumulative risk of PPD in both relatives of AD probands with and without the epsilon 4 allele was significantly higher than that in the relatives of non-demented controls. However, the increased risk in the relatives of AD probands with the epsilon 4 allele was marginally, but not significantly, lower than the risk in the relatives of probands without epsilon 4. A greater likelihood of death by heart diseases over developing PPD in relatives of AD probands with epsilon 4 (3.1-fold increase) was found compared to relatives of probands without epsilon 4 (1.7-fold increase), especially prior to age 70, although the difference was not statistically significant. The increased familial risk for PPD in the relatives of AD probands with the APOE-epsilon 4 allele relative to controls suggests that familial factors in addition to APOE-epsilon 4 are risk factors for AD. Differential censorship from increased mortality of heart diseases may have prevented a higher incidence of PPD among the relatives of probands with epsilon 4.