This study examined subgroup differences in the effectiveness of a universal classroom-based preventive intervention. The Good Behavior Game (GBG) was delivered in Grade 1 and 2 in a randomized controlled trial including 759 students. Changes in externalizing and internalizing problems were modeled from Kindergarten through Grade 2. Unlike previous research, a person-centered approach was employed to examine critical combinations of child, peer, family, and demographic characteristics at baseline as moderators of intervention impact. Six subgroups were identified that differed both in baseline risk profiles and intervention responsiveness. The GBG prevented the development of externalizing and internalizing behavior among low-risk children, children with emotional problems, and victimized children. No positive intervention effects were found for children from dysfunctional families and children with combinations of behavioral and social risks. The study presented a novel approach to study subgroup differences in universal preventive interventions and provides first evidence that universal school-based programs may not be effective for children with more severe risks and risks at multiple levels. © 2013 Society for Prevention Research.