Ratings of child involvement in manual-based cognitive-behavioral treatment for anxiety were associated with the absence of primary anxiety diagnosis and reductions in impairment ratings at posttreatment for 59 children with anxiety (ages 8-14 years). Good-to-excellent interrater reliability was established for the independent ratings of 237 therapy sessions, and strong psychometrics were established for the involvement measure. Child involvement at midtreatment, just prior to in vivo exposures, was positively associated with treatment gains but earlier involvement was not. Increased involvement during therapy (positive involvement shifts) may provide a useful index of change and may also predict outcomes. Involvement was not associated with client demographics or diagnostic category. Implications for treatment and measurement of psychotherapy process within manual-based treatments are discussed.