In an elderly population of 4,932 that included 694 subjects aged 85 and older, we studied the occurrence of Alzheimer disease (AD) in relation to APOE genotype1, 2. APOE has three alleles, epsilon2, epsilon3 and epsilon4 — the latter of which has been associated with an increased age-specific risk of AD (ref. 1). We screened all subjects for cognitive dysfunction using either the modified mini-mental state examination3 (3MS) or a structured questionnaire4. A weighted random sample of 878 subjects was then studied with both the dementia questionnaire5 (DQ) and a detailed follow-up clinical assessment6 (CA). We investigated the remaining 4,054 with the DQ and CA when their prior results suggested cognitive impairment. A physician (D.C.S. or J.C.S.B.) reviewed the results and examined most subjects with apparent dementia. Medical records and laboratory studies (including neuroimaging) were sought, and a consensus panel of geropsychiatrists, neurologists and neuropsychologists then made final diagnoses and assigned onset ages, defined as the year subjects unambiguously met DSM-III-R criteria for dementia. Onset ages were believed accurate to within a year in most cases. We found 322 demented individuals, 220 with definite, probable or possible AD (ref. 7). In 25 autopsied cases, 17 of 18 clinical diagnoses of probable or possible AD (94.4%) were confirmed neuropathologically. Four of the five other subjects with AD neuropathology had been categorized clinically as, "dementia, undetermined aetiology," which retains AD in the differential diagnosis.