The correlation between incidence of senile dementia and variables determined in baseline health examinations was analyzed in a retrospective cohort study. The study cohort was assembled in 1965 and followed up to 1985. A survey of dementia was performed on those alive in 1985, and on the decreased through their families who were queried about his or her condition just before death. The survey was conducted by public health nurses and assistants. The results obtained were as follows: 1. There was a slightly higher incidence in women and incidence for both sexes demonstrated an increase with age. 2. Analysis by cause of death showed increased death by cerebrovascular diseases for both men and women, and by cardiovascular disease for women, but decreased death from cancer. 3. Correlations of senile dementia with high blood pressure and albuminuria findings at the initial examination were seen. In addition, results of fundus oculi examination in men, and electrocardiography in women were also regarded as risk factors for the occurrence of senile dementia. 4. Comparative analysis showed higher incidence among farmers rather than fishermen, and in drinkers rather than nondrinkers, while no relation was found to smoking habits.