Acutely ill patients presenting for admission in two district psychiatric services were randomly allocated to day-hospital or in-patient care. In both services a quarter of all admissions could not be allocated because they were too ill (half of these were compulsory admissions); these patients were predominantly manic and schizophrenic patients with pronounced psychotic symptoms and disturbed behaviour. In one service 80% (41/51) of patients randomly allocated to day-hospital treatment were successfully engaged in treatment compared with 54% (19/35) in the other service. This difference arose because only patients with mildly disturbed behaviour could be treated at the second day hospital. For patients who were successfully allocated, the outcome of treatment was similar in terms of psychiatric symptoms and social functioning. The results of a treatment trial for acutely ill patients in one district service cannot be generalised to other district services without due attention to the factors, such as staffing levels, which determine the degree of disturbed behaviour that can be tolerated in the day hospital.