Few investigators studying alcohol abuse among individuals with a severe mental illness (SMI) have examined predictors of posttreatment alcohol outcomes. In the present study, a multivariate approach based on a theoretical model was used to study the relationship between psychosocial factors and post-treatment-initiation alcohol use. Predictors of alcohol use outcomes were examined in 278 individuals diagnosed with a current schizophrenia-spectrum or bipolar disorder and an alcohol use disorder (AUD). At 6-months follow-up, 144 of 228 available participants (63%) had good clinical outcomes. The results of structural equation modeling indicated that type of pretreatment residential setting was directly related to treatment, with participants who lived in supervised settings (41%) reporting significantly more days of treatment ([beta] = 0.34, p less than 0.001). In addition, participants with more psychiatric symptoms, as assessed with the Brief Symptom Inventory and Structured Clinical Interview for the Positive and Negative Syndrome Scale, reported significantly fewer treatment days ([beta] = -0.20, p less than 0.001). Number of days that participants attended treatment was indirectly associated with alcohol use outcomes and was mediated by use of alcohol-specific coping skills, such that more frequent use of such skills was associated with less post-treatment-initiation alcohol use ([beta] = -0.34, p less than 0.001). This study emphasizes the favorable prognosis for alcohol outcomes among treated individuals with SMI and AUD and the importance of psychosocial interventions, particularly those that result in better alcohol-specific coping skills. (Contains 4 figures, 2 tables and 2 footnotes.)