Whereas early-onset Alzheimer disease (AD; usually onset at age < 50 years) has been defined with genetic mutation on chromosomes 1, 14, and 21, the degree of familial contribution to late-onset AD is unclear. Further, it is uncertain if subgroups of late-onset AD exist. To examine the influence of familial factors as a function of age in late-onset AD we investigated lifetime risks and age-specific hazard rates of AD-like illness among late-onset AD probands' and controls' first-degree relatives, using questionnaires and medical records. As part of a longitudinal study on aging and AD, we studied 78 AD probands with age of onset > or =50 years (28 "definite" and 50 "probable" AD according to NINCDS/ADRDA criteria) and 101 healthy old controls seen since 1981. Both probands and controls were screened rigorously with medical tests and brain imaging and seen regularly until autopsy. Multiple informants and medical records were used for first-degree relatives. Among first-degree relatives, 49 secondary cases of AD-like illness were found for the AD probands' relatives (391 relatives 40 years old or older) compared with 20 cases among controls' relatives (456 relatives 40 years old or older). Relatives of AD probands had a significantly increased lifetime risk of AD-like illness of 52.8+/-11.4% by age 94 years compared with a lifetime risk in relatives of controls of 22.1+/-5.8% by age 90 years. Age-specific hazard rates in relatives of AD probands increased until the 75-79-year age interval and then decreased; in contrast the age-specific hazard rates increased in relatives of controls after the 80-84-year age interval. To determine if a dividing line exist among late-onset AD, several cutoff ages were used in our study to compare cumulative risk curves of AD-like illness between relatives of late-onset probands and relatives of late-late-onset probands. Differences in the pattern of cumulative incidence of AD in relatives showed that 67-71 years is the range for a dividing line between late- and late-late-onset AD. Age-specific hazard rates of AD in relatives supported a difference between late- and late-late-onset. Whereas these rates increased until the 75-79-year age interval and then decreased in late-onset AD, the rates began increasing after the 65-69-year age interval and through the oldest age interval in both late-late-onset AD and control groups. Our results support the concept that familial factors exist in late-onset AD and that different familial factors may exist in late-onset AD subgroups.