OBJECTIVE: To evaluate the effectiveness of a crisis resolution team. DESIGN: Randomised controlled trial. PARTICIPANTS: 260 residents of the inner London Borough of Islington who were experiencing crises severe enough for hospital admission to be considered. INTERVENTIONS: Acute care including a 24 hour crisis resolution team (experimental group), compared with standard care from inpatient services and community mental health teams (control group). MAIN OUTCOME MEASURES: Hospital admission and patients' satisfaction. RESULTS: Patients in the experimental group were less likely to be admitted to hospital in the eight weeks after the crisis (odds ratio 0.19, 95% confidence interval 0.11 to 0.32), though compulsory admission was not significantly reduced. A difference of 1.6 points in the mean score on the client satisfaction questionnaire (CSQ-8) was not quite significant (P = 0.07), although it became so after adjustment for baseline characteristics (P = 0.002). CONCLUSION: Crisis resolution teams can reduce hospital admissions in mental health crises. They may also increase satisfaction in patients, but this was an equivocal finding.