Objectives: to evaluate the effect of a care home rehabilitation service on institutionalisation, health outcomes and service use. Design: randomised controlled trial, stratified by Barthel ADL index, social service sector and whether living alone. The intervention was a rehabilitation service based in Social Services old people's homes in Nottingham, UK. The control group received usual health and social care. Participants: 165 elderly and disabled hospitalised patients who wished to go home but were at high risk of institutionalisation (81 intervention, 84 control). Main outcome measures: institutionalisation rates, Barthel ADL index, Nottingham Extended ADL score, General Health Questionnaire (12 item version) at 3 and 12 months, Health and Social Service resource use. Results: the number of participants institutionalised was similar at 3 months (relative risk 1.04, 95% confidence intervals 0.65–1.65) and 12 months (relative risk 1.23, 95% confidence intervals 0.75–2.02). Barthel ADL Index, Nottingham Extended ADL score and General Health Questionnaire scores were similar at 3 and 12 months. The intervention group spent significantly fewer days in hospital over 3 and 12 months (mean reduction 12.1 and 27.6 days respectively, P