Alcohol-dependent patients (N = 97) were randomly assigned to receive either naltrexone or placebo and either relapse prevention therapy or supportive therapy. The present report explored the hypothesis that patients could be matched to the above treatments on the basis of specific pretreatment characteristics. Treatment matching variables explored included craving, alcohol dependence severity, and cognitive measures of learning and memory. Results of linear regression analyses tentatively suggest that patients experiencing higher levels of craving and poorer cognitive functioning may derive the greatest benefit from naltrexone versus placebo. For psychotherapy, lower levels of verbal learning were associated with poorer drinking outcomes for relapse prevention therapy but not for supportive therapy. Conversely, higher levels of verbal learning were associated with better outcomes for relapse prevention therapy but not for supportive therapy.