Objectives: To investigate the association between midlife risk factors and the development of vascular dementia (VaD) or Alzheimer's disease (AD) 25 to 30 years later.

Design: A prevalence study within a longitudinal cohort study.

Setting: Subjects in the Adult Health Study (a prospective cohort study begun in 1958) have been followed through biennial medical examinations in Hiroshima, Japan.

Participants: One thousand seven hundred seventy-four subjects in Hiroshima, Japan born before September 1932 (1,660 with no dementia, 114 with dementia (51 with AD, and 38 with VaD) diagnosed from 1992 to 1997 according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria).

Measurements: The subjects were examined for effect on dementia of sex, age, education, atomic bomb radiation dose, and midlife factors associated with risk (smoking, alcohol intake, physical activity, dietary habits, systolic blood pressure (SBP), body mass index, and history of diabetes mellitus) that had been evaluated in 1965-1970.

Results: VaD prevalence increased significantly with age, higher SBP, and lower milk intake. The odds ratios of VaD for age (in 5-year increments), SBP (10 mmHg increments), and milk intake (almost daily/less than four times a week) were 1.29, 1.33, and 0.35, respectively. The risk factors for VaD were compatible with the risk factors for stroke in this study population. AD prevalence increased significantly with age and lower education. Other midlife factors and radiation dose did not show any significant association with VaD or AD.

Conclusion: Increased SBP and low milk intake in midlife were associated with VaD detected 25 to 30 years later. Early behavioral control of the risk factors for vascular disease might reduce the risk of dementia.