The costs of home care and treatment solely in hospital for patients in a variety of short-term diagnostic categories are compared. Five hundred and eighty-three patients included in an experimental home care program were randomly assigned either to a group which received home care as part of their treatment, or to a control group that remained in hospital the traditional length of time. It is argued that the only costs relevant in an economic comparison of the two modes of treatment are those attributable to the direct care of the patient. A technique is presented whereby changes in the daily amount of nursing service provided can be costed. The economic analysis shows that, when similar diagnoses are compared for an episode of illness, there is basically no difference in cost between home care and treatment in hospital.