Groups of middle-income outpatient problem drinkiers were randomly assigned to one of two treatments: a multicomponent positive reinforcement procedure emphasizing moderation, or "traditional" denial-confrontation therapy emphasizing abstinence. There were no significant differences between participants assigned to the two procedures with respect to pretreatment characteristics. Participants reduced their alcohol consumption significantly in both treatments. Behavioral participants reduced alcohol consumption significantly during therapy, whereas traditional participants reduced significantly prior to therapy. Significantly more traditional participants dropped out of therapy than did behavioral. The data suggested that the effects of traditional therapy were mixed, helping those participants who were receptive, driving out those who were not; moreover, reduced consumption seemed to be more the result of compliance with initial therapist demands (made during the screening period) than the effect of denial-confrontation in subsequent therapy. Behavioral treatment was more successful in several indicators of outcome; the improvement observed seemed to be attributable to the active therapeutic techniques employed.