Antioxidants have been hypothesized to protect against Alzheimer's disease, but studies conducted in late life have been inconsistent. Risk factors measured in midlife may better predict dementia in late life because they are less affected by the disease process. The authors examined the association of midlife dietary intake of antioxidants to late-life dementia and its subtypes. Data were obtained from the Honolulu-Asia Aging Study, a prospective community-based study of Japanese-American men who were aged 45-68 years in 1965-1968, when a 24-hour dietary recall was administered. The analysis included 2,459 men with complete dietary data who were dementia-free at the first assessment in 1991-1993 and were examined up to two times for dementia between 1991 and 1999. The sample included 235 incident cases of dementia (102 cases of Alzheimer's disease, 38 cases of Alzheimer's disease with contributing cerebrovascular disease, and 44 cases of vascular dementia). Relative risks by quartile of intake were calculated using Cox proportional hazards models with age as the time scale, after adjustment for sociodemographic and lifestyle factors, cardiovascular risk factors, other dietary constituents, and apolipoprotein E e4. Intakes of beta-carotene, flavonoids, and vitamins E and C were not associated with the risk of dementia or its subtypes. This analysis suggests that midlife dietary intake of antioxidants does not modify the risk of late-life dementia or its most prevalent subtypes.