Using receiver operating characteristics, the authors examined outcome predictors (variables associated with outcome regardless of treatment) and moderators (variables identifying subgroups with differential treatment effectiveness) in the Multimodal Treatment Study of Children with Attention-Deficit/Hyperactivity Disorder (ADHD; MTA). Treatment response was determined using parent- and teacher-reported ADHD and oppositional defiant symptoms, with levels near or within the normal range indicating excellent response. Among 9 baseline child and family characteristics, none predicted but 3 moderated treatment response. In medication management and combined treatments, parental depressive symptoms and severity of child ADHD were associated with decreased rates of excellent response; when these 2 characteristics were present, below-average child IQ was an additional moderator. No predictors or moderators emerged for behavioral and community comparison treatments. The authors discuss conceptual and clinical implications of research on treatment moderators.