The purpose of this study was to establish the predictive validity of guidelines for allocating patients to outpatient or inpatient treatment for an alcohol-use disorder. It was hypothesized that patients who were matched to the recommended level of care would have (a) better outcomes than patients treated at a less intensive level of care, and (b) outcomes equivalent to those of patients treated at a more intensive level of care. Matched patients were allocated according to an algorithm based on their treatment history, addiction severity, psychiatric impairment, and social stability at baseline. Outcome was measured in terms of self-reported alcohol use 30. days prior to follow-up and changes in number of abstinent and heavy drinking days between intake and follow up. Of the 2,310 patients, 65.4% were successfully followed up 9.67. months after intake. Only 22% of the patients were treated according to the level of care prescribed by the guidelines; 49% were undertreated; and 29% were overtreated. The results were not in line with our hypotheses. Patients treated at a more intensive level of care than recommended had favorable outcomes compared to patients treated at the recommended level of care (55.5% vs. 43.9% success). Patients allocated to the recommended level of care did not have better outcomes than those treated at a less intensive level of care (43.9% vs. 38.3% success). Based on these results, we suggest ways to improve the algorithm for allocating patients to treatment. © 2012 Elsevier Ltd.