Patient treatment matching hypotheses were tested for substance users randomly assigned to a group cognitive behavioral therapy (CBT; n= 114) or a group motivational intervention (GMI; n= 116). Treatment was scheduled twice weekly for 10 weeks. Using a patient attribute by treatment interaction design with a 15-week follow-up, the study predicted that alexithymia, antisocial personality disorder (ASPD), and network support for alcohol and drug use would be associated with less substance use for CBT subjects and that hostility and lower treatment motivation would be associated with less substance use for GMI subjects. Three of the hypothesized moderators were empirically supported: alexithymia, network support for alcohol, and ASPD. Results indicate the use of assessing specific patient attributes to better inform treatment recommendations.