OBJECTIVES: to evaluate the outcome and cost of transfer to a nursing-led inpatient unit for ‘intermediate care’. The unit was designed to replace a period of care in acute hospital wards and promote recovery before discharge to the community. DESIGN: randomized controlled trial comparing outcomes of care on a nursing-led inpatient unit with the system of consultant-managed care on a range of acute hospital wards. SETTING: hospital wards in an acute inner-London National Health Service trust. SUBJECTS: 175 patients assessed to be medically stable but requiring further inpatient care, referred to the unit from acute wards. INTERVENTION: 89 patients were randomly allocated to care on the unit (nursing-led care with no routine medical intervention) and 86 to usual hospital care. MAIN OUTCOME MEASURES: length of hospital stay, discharge destination, functional dependence (Barthel index) and direct healthcare costs. RESULTS: care in the unit had no significant impact on discharge destination or dependence. Length of inpatient stay was significantly increased for the treatment group (P=0.036; 95% confidence interval 1.1–20.7 days). The daily cost of care was lower on the unit, but the mean total cost was £1044 higher—although the difference from the control was not significant (P=0.150; 95% confidence interval -£382 to £2471). CONCLUSIONS: the nursing-led inpatient unit led to longer hospital stays. Since length of stay is the main driver of costs, this model of care—at least as implemented here—may be more costly. However, since the unit may substitute for both secondary and primary care, longer-term follow-up is needed to determine whether patients are better prepared for discharge under this model of care, resulting in reduced primary-care costs.