The study's aim was to investigate the cost-effectiveness of an NHS/Social Services short-term residential rehabilitation unit (a form of intermediate care) for older people on discharge from community hospital compared with 'usual' community services. An economic evaluation was conducted alongside a prospective controlled trial, which explored the effectiveness of a rehabilitation unit in a practice setting. The aim of the unit was to help individuals regain independence. A matched control group went home from hospital with the health/social care services they would ordinarily receive. The research was conducted in two matched geographical areas in Devon: one with a rehabilitation unit, one without. Participants were recruited from January 1999 to October 2000 in 10 community hospitals and their eligibility determined using the unit's strict inclusion/exclusion criteria, including 55 years or older and likely to benefit from a short-term rehabilitation programme: potential to improve, realistic, achievable goals, motivation to participate. Ninety-four people were recruited to the intervention and 112 to the control group. Details were collated of the NHS and Social Services resources participants used over a 12-month follow-up. The cost of the resource use was compared between those who went to the unit and those who went straight home. Overall, costs were very similar between the two groups. Aggregated mean NHS/Social Services costs for the 12 months of follow-up were pound 8542.28 for the intervention group and pound 8510.68 for the control. However, there was a clear 'seesaw' effect between the NHS and Social Services: the cost of the unit option fell more heavily on Social Services (pound 5011.56, whereas pound 3530.72 to the NHS), the community option more so on the NHS (pound 5146.74, whereas pound 3363.94 to Social Services). This suggests that residential rehabilitation for older people is no more cost-effective over a year after discharge from community hospital than usual community services. The variability in cost burden between the NHS and Social Services has implications for 'who pays' and being sure that agencies share both pain and gain.