Epidemiological surveys of senile dementia at home were carried out on 3,754 subjects for two consecutive years, 1987 and 1988. The prevalence and incidence rates were 4.1% and 1.01%, respectively. Concerning the incidence rate, the sample populations of previous reports were small in number--less than 1,000. If we suppose that the incidence were to be approximately 1%, then it follows that dementia would be noted yearly in no more than one out of 100 senile subjects. A difference in subject number leads to a great difference in the incidence. Therefore, it is necessary that the number of subjects should be more than 1,000 for a comprehensive epidemiological survey of incidence rate of senile dementia. Accordingly, the present study is a better representation of the true picture as it was obtained by a more accurate method accompanied by less errors.