Aim To test a priori hypotheses concerning client–treatment matching in the treatment of alcohol problems and to evaluate the more general hypothesis that client–treatment matching adds to the overall effectiveness of treatment. Design Pragmatic, multi-centre, randomized controlled trial (the UK Alcohol Treatment Trial: UKATT) with open follow-up at 3 months after entry and blind follow-up at 12 months. Setting Five treatment centres, comprising seven treatment sites, including National Health Service (NHS), social services and joint NHS/non-statutory facilities. Treatments Motivational enhancement therapy and social behaviour and network therapy. Measurements Matching hypotheses were tested by examining interactions between client attributes and treatment types at both 3 and 12 months follow-up using the outcome variables of percentage days abstinent, drinks per drinking day and scores on the Alcohol Problems Questionnaire and Leeds Dependence Questionnaire. Findings None of five matching hypotheses was confirmed at either follow-up point on any outcome variable. Conclusion The findings strongly support the conclusion reached in Project MATCH in the United States that client–treatment matching, at least of the kind examined, is unlikely to result in substantial improvements to the effectiveness of treatment for alcohol problems. Possible reasons for this failure to support the general matching hypothesis are discussed, as are the implications of UKATT findings for the provision of treatment for alcohol problems in the United Kingdom.