Ninety-six subjects were recruited from an inpatient alcoholism treatment program and randomly assigned to aftercare group treatment with either coping skills training or interactional therapy. Survival analyses using 2-year outcome data provided evidence for the durability of matching interaction effects. The data supported a priori hypotheses that individuals scoring high on measures of sociopathy or global psychopathology have better outcomes in coping skills treatment, whereas patients low on these dimensions have better outcomes in interactional treatment. Contrary to the original hypothesis, patients with cognitive impairment had better outcomes in interactional treatment, and patients without cognitive impairment did better in coping skills treatment.