Objective: In this study, we evaluated parent and child characteristics as predictors and moderators of response in the four-site Treatment of Severe Childhood Aggression (TOSCA) study. Methods: A total of 168 children with severe aggression, disruptive behavior disorder, and attention-deficit/hyperactivity disorder (ADHD) were enrolled in a 9-week trial of basic treatment (n = 84, stimulant + parent training + placebo) versus augmented treatment (n = 84, stimulant + parent training + risperidone). In the initial report, augmented treatment surpassed basic treatment in reducing the primary outcome of disruptive behavior (D-Total) scores. In the current study, we evaluated parent (income, education, family functioning, employment) and child variables (intelligence quotient [IQ], aggression type, comorbid symptomatology) as predictors or moderators, using linear mixed models and the MacArthur guidelines. Results: Higher scores on ADHD symptom severity and callous/unemotional traits predicted better outcome on D-Total regardless of treatment assignment. Two moderators of D-Total were found: Higher anger/irritability symptoms and lower mania scores were associated with faster response, although not better overall effect at endpoint, in the augmented but not the basic group. Several variables moderated response on secondary outcomes (ADHD severity and prosocial behavior), and were characterized by faster response, although not better outcome, in the augmented but not in the basic group. Maternal education moderated outcome on the measure of positive social behavior; children of mothers with less education benefited more from augmented treatment relative to basic than those with more education. Conclusion: Although these findings require validation, they tentatively suggest that augmented treatment works equally well across the entire sample. Nevertheless, certain child characteristics may be useful indicators for the speed of response to augmented treatment.