The objectives of this study were to explore the costs and outcomes associated with different types of intermediate care (IC) services, and also to examine the characteristics of patients receiving such services. Five UK case studies of 'whole systems' of IC were used, with data collected on a sample of consecutive IC episodes between January 2003 and January 2004. Statistical differences in costs and outcomes associated with different IC services and patient groups were explored. Factors associated with variation in IC episode outcomes (EuroQol EQ-5D and Barthel Index) were explored using an econometric framework. Data were available for 2253 episodes of IC. In terms of Department of Health criteria, a large proportion of patients (up to 47% of those for whom data were available) in this study were inappropriately admitted to IC services. As regards service function, compared to supported discharge, admission avoidance services were associated with both lower costs and greater health and functional gains. These gains appear to be driven, in part, by illness severity (more dependent patients tended to gain most benefit). In addition, these gains appear to be larger where the admission was appropriate. Our work suggests a need for the development and application of robust and reliable clinical criteria for admission to IC, and close co-operation between hospital and community service providers over selection of patients and targeting of IC and acute care services to meet defined clinical need.