This study examined predictive relations between 9 therapist behaviors and client involvement in manual-guided, cognitive-behavioral therapy for adolescent depression. Analyses included 42 adolescents who met criteria for a depressive disorder (major depressive disorder, dysthymic disorder, or adjustment disorder with depressed mood) and who were treated in school-based clinics. Therapist behaviors hypothesized to promote client involvement were coded from Session 1 audiotapes; client involvement was coded from Session 2. Unlike prior research, the current study examined associations between behaviors and involvement while controlling for initial client resistance to isolate the therapist contribution to involvement. Results show that 3 therapist behaviors from Session 1 (attending to teen's experience, exploring teen's motivation, and less structure) predicted greater client involvement in Session 2, controlling for initial resistance. Only exploring motivation and less structure uniquely predicted Session 2 involvement when the 3 behaviors were examined simultaneously. Session 1 therapist behaviors predicted significant variance in involvement at Sessions 2, 4, and 8. Client initial presentation as resistant was associated with more exploring motivation and praising, but initial resistance did not explain associations between therapist behaviors and involvement. Implications for implementing evidence-based treatments are discussed. (Contains 2 tables and 1 footnote.)