To estimate the risk factors for intellectual dysfunction and examine its prognosis in a community-residing (non-institutionalized) elderly population, a randomly selected sample of 1,473 elderly people aged 65 years and over living in S city, Osaka Prefecture, was studied in October 1992, and data were obtained from 1,383, a response rate of 93.9%. A cohort of 1,383 was followed for 42 months and follow-up was completed for 1,300 (94.0%). The main results were as follows: 1) The prevalence of intellectual dysfunction did not differ significantly between sexes, and there was an increasing prevalence of intellectual dysfunction with age in both sexes. The prevalence of severe intellectual dysfunction was found to increase highly at age 85 and over. 2) By univariate analysis, odds ratios for age older than 75 years, low Activities of Daily Living (ADL), urinary and fecal incontinence, and no participation in social activities were significantly higher than 1 in any level of mild, moderate, and severe intellectual dysfunction. In the multivariate analysis using logistic regression, age older than 75 years and urinary and fecal incontinence showed significant higher odds ratios than 1 for severe intellectual dysfunction, and low ADL and treatment for hypertension also showed significant higher odds ratios than 1 for moderate intellectual dysfunction. 3) From analysis using the Kaplan-Meier method, the cumulative survival rates decreased with a decline in intellectual functioning in both age groups of 65-74 and 75 years and older. 4) Application of the Cox proportional hazards model resulted in adjusted hazard ratio for severe intellectual dysfunction of 1.79 (95% confidence interval, 1.02-3.12), controlling for other factors such as sex, age, general health status, incontinence and social activities.