The purpose of this study was to determine lifestyle factors in the elderly that affected longevity, using a population-based prospective study. The participants were 440 men and 625 women aged 60 to 74 living in a rural Japanese community. The baseline data such as age, sex, present illness, walking hours per day, sleeping hours per day, alcohol consumption, a history of smoking, and "ikigai" (meaningfulness of life) were collected in July 1990. During 90 months of follow-up from July 1990 to December 31 1997, there were 123 deaths. By Cox's multivariate hazard model adjusted age, sex, and medical histories, walking > or = 1 hour/day (HR = 0.63, 95% CI 0.44-0.91) and an "ikigai" (HR = 0.66, 95% CI 0.44-0.99) lowered the risk for all-cause mortality independently. In regard to hours of sleep, the cumulative survival curve showed that 7 hours/day was the border and sleeping > or = 7 hours/day lowered the risk (HR = 0.49 95% CI 0.33-0.74). Based on the findings in this study, walking > or = 1 hour/day, sleeping > or = 7 hours/day, and "ikigai" are important factors for longevity in the elderly.