OBJECTIVE: To examine the relationship between incontinence and mortality in elderly people living at home. DESIGN: Of the randomly selected people aged 65 years and older living in Settsu city, Osaka in October 1992, 1405 were contacted and constituted the study cohort. Follow-up for 42 months was completed for 1318 (93.8%; 1129 alive, 189 dead). MEASURES: Data on general health status, history of health management, psychosocial conditions and urinary and faecal incontinence were collected by interview during home visits at the time of enrolment. RESULTS: From the Kaplan-Meier analysis, the estimated survival rates decreased with a decline in continence in both the 65-74 and 75 years and older age groups. From the Cox proportional hazards model, unadjusted hazard ratios of minor, moderate and severe incontinence for mortality, compared with continence, were 2.27, 2.96 and 5.94, respectively. Multivariate analysis yielded adjusted hazard ratios of minor, moderate and severe incontinence of 0.99, 1.17 and 1.91, respectively, leaving severe incontinence as the significant factor, when other indicators are controlled. CONCLUSIONS: Incontinence is related to mortality and severe incontinence represents an increased risk factor for mortality in elderly people living at home.