Objective: To identify predictors and moderators of response to acute treatments among depressed adolescents (N = 439) randomly assigned to fluoxetine, cognitive-behavioral therapy (CBT), both fluoxetine and CBT, or clinical management with pill placebo in the Treatment for Adolescents With Depression Study (TADS). Method: Potential baseline predictors and moderators were identified by a literature review. The outcome measure was a week 12 predicted score derived from the Children's Depression Rating Scale-Revised (CDRS-R). For each candidate moderator or predictor, a general linear model was conducted to examine main and interactive effects of treatment and the candidate variable on the CDRS-R predicted scores. Results: Adolescents who were younger, less chronically depressed, higher functioning, and less hopeless with less suicidal ideation, fewer melancholic features or comorbid diagnoses, and greater expectations for improvement were more likely to benefit acutely than their counterparts. Combined treatment, under no condition less effective than monotherapy, was more effective than fluoxetine for mild to moderate depression and for depression with high levels of cognitive distortion, but not for severe depression or depression with low levels of cognitive distortion. Adolescents from high-income families were as likely to benefit from CBT alone as from combined treatment. Conclusions: Younger and less severely impaired adolescents are likely to respond better to acute treatment than older, more impaired, or multiply comorbid adolescents. Family income level, cognitive distortions, and severity of depression may help clinicians to choose among acute interventions, but combined treatment proved robust in the presence of moderators. (Contains 1 figure and 5 tables.)