While research has shown community-based psychiatric care to be as good as, or better than, hospital-based care, generalisation to clinical practice has been difficult. This prospective, randomised controlled study examined a community-based approach feasible within NHS conditions. Ninety-four patients were randomly allocated to experimental and 78 to control treatments and followed for one year. The groups were well matched apart from an excess of psychotic control patients. No differences in clinical or social functioning outcome were found. Both groups improved substantially on clinical measures in the first six weeks, with some slow consolidation thereafter. There were three suicides in the control group and one in the experimental group. Access to care was better in the experimental group (93% attended assessment) than in the control group (75% attended assessment).