This study examined two issues related to the distinction between unitary and multidimensional models of alcoholism treatment outcome. First, does treatment outcome vary along a variety of relatively independent dimensions? Second, how important is abstinence to improved health and psychosocial adjustment? A sample of 266 alcoholics were evaluated at intake and 1 year following inpatient treatment using an extensive battery of assessments. Correlational analysis and factor analysis provided limited support for both the unidimensional and multidimensional approaches. A synthetic measure of posttreatment drinking showed a clear linear relationship between level of consumption and lack of improvement in medical status, biological function, life stress and psychopathology. The findings are discussed in terms of the methodological difficulties in treatment evaluation and the need for more systematic research on the effect of posttreatment drinking on multiple dimensions of outcome status.