High rates of mortality are an abiding feature of homes for the elderly. Almost a fifth (19.2%) of a cohort of 6947 residents died in the 12 months following intial assessment. Survival rates vary widely, however, for different groups within the residential population. Using data from a longitudinal study of 175 homes for the elderly, this paper examines the relationship between mortality, age, length of stay, and dependency. The interaction between these variables within the residential setting helps to identify the ‘high risk’ groups of residents who require special surveillance and care.