A cognitive-behavioral program with a goal of either abstinence or controlled drinking was assessed. The 70 early-stage problem drinkers were randomly assigned to one of the two goal conditions, and within each condition to one of two therapists. On average they received six individual weekly sessions, each lasting approximately 90 min. Both groups were taught to indentify risk situations and existing competencies, to develop cognitive and behavioral coping, and to assess their progress objectively. The controlled-drinking group was also taught procedures for moderate drinking. Over the 2-year follow-up period, no significant differences were found between the groups in reported alcohol consumption. Six months after treatment drinking had been reduced from an average of about 51 drinks per week to 13, and this reduction was maintained throughout the second year. Reports of drinking were corroborated by independent measures. Although the outcomes of the groups were similar, controlled drinking was considered to be a more suitable goal; it was more acceptable to the majority of the clients, and most of those assigned to abstinence developed moderate drinking on their own. The authors wish to thank the following colleagues for their contributions to the project: C. Milton-Feasby and V. Ittig-Deland for conducting intake and follow-up assessments, B. Sisson for medical screenings, Y. Israel for recommending the biochemical corroborators of alcohol consumption, and B. Kapur for conducting the laboratory analyses. Special appreciation is extended to H. Lei for his advice on statistical analyses and to D. A. Wilkinson, K. D. Walker, H. A. Skinner, and C. Poulos for their valuable feedback and support.Requests for reprints should be sent to M.