The objective of this study was to compare hospitalisation rates by cause of admission, hospital death rates and length of stay for residents from nursing and residential care homes with those in the community. This is a retrospective study of acute hospital emergency admissions in one health district, Merton, Sutton and Wandsworth between April 1996 and March 1997. Data linkage and manual look up were used to derive emergency hospital admissions for residents of care homes aged 65 and over. Admission rates were calculated for cause, length of stay and hospital death for residents of care homes and in the community with relative risks. The relative risk of emergency admission from a care home compared with the community was 1.39 for all diagnoses, 2.68 for all injuries, and 3.96 for fracture of neck of femur. The relative risk of dying in hospital for care home residents was 2.58 overall, and 3.64 in the first 48 hours of a hospital stay (all P-values <0.0001). Admission rates were higher from residential than from nursing homes. There was some increase in admissions from homes during holiday periods and over Christmas. In conclusion, there are major difficulties in monitoring admissions from nursing and residential care homes due to poor quality recording and inaccuracies in NHS coding. This was compounded by an absence of data on the age and sex profile and healthcare needs of the resident population in care homes. Prospective studies are required to ascertain when admission is avoidable and when it is appropriate. The information strategy needs to ensure that routine data sources are capable of monitoring the use of hospital services by residents of care homes.