The aim of this study was to examine the life expectancy of elderly people in nursing and residential care over a 20-month period and its relationship to specific risk factors. Using a retrospective cohort design, data obtained on 1888 residents placed between 1 July 1997 and 30 April 1999 in residential, nursing and dual registered homes within Nottingham Health Authority boundaries were examined. Additional data on physical and mental disability at placement were available for 514 residents. Main outcome measures comprised survival rate overall, and in relation to gender, age, home type (nursing, residential or dual), source of placement (hospital or community) and various disability factors.One-year survival rates were: overall, 66%; nursing homes, 59%; dual homes, 58%; and residential homes 76%. Median survival in nursing homes was 541 days, but was not reached in residential homes. Male gender, admission to nursing or dual registered homes, placement from hospital, decreased mobility and increased age were associated with decreased life expectancy. Although no association was found between length of survival and level of cognitive function, lack of cognitive impairment was associated with lower survival. In conclusion, mortality is high in nursing, dual and residential homes where life expectancy has been shown to be associated with gender, home type, origin of placement and mobility. Rates of survival are related to higher comorbidity and disability. Important data for planning and assessing care needs can be yielded through the analysis of mortality data.