An accurate population- based prospective cohort study of dementia was begun in 1985 in the town of Hisayama, Japan. To examine secular trends in the prevalence of dementia, four cross- sectional surveys were conducted among Hisayama residents aged 65 years or older in 1985, 1992, 1998, and 2005. The prevalence of Alzheimer’s disease (AD) significantly increased with time, while the prevalence of vascular dementia (VaD) decreased in the early period and then increased in late period. In a cohort study of incidence of dementia among Hisayama elder residents without dementia, the survival curve of dementia cases aged 65 to 89 years was significantly lower than that of age- and sex- matched control subjects randomly selected from those without incident dementia. We demonstrated that late- life hypertension was a significant risk factor for the development of VaD but not for AD, and people with mid- life hypertension were at an increasing risk of late- life onset of VaD regardless of late- life hypertension status. Meanwhile, we revealed the significant associations between glucose intolerance/diabetes and the risks of developing both VaD and AD. Additionally, elevating 2- hour postload glucose levels were closely associated with increased risk of VaD and AD.