The current study examined the relationship between child involvement and treatment outcome in a group cognitive–behavioral treatment (GCBT) program for children with anxiety, using an adapted version of the Child Involvement Rating Scale (CIRS). Thirty-four children with a primary diagnosis of an anxiety disorder, referred for outpatient treatment at a hospital, and their parents participated in a manualized 12-session GCBT program. The Multidimensional Anxiety Scale for Children (MASC), administered at pretreatment, posttreatment, and at 4-month follow-up, served as the outcome measure. The psychometric properties of the adapted CIRS were good. Early involvement (EI) was calculated as the average score from Sessions 1 to 6 and late involvement (LI), the average score from Sessions 7 to 12. EI predicted lower MASC scores at posttreatment, after controlling for pretreatment MASC scores and for LI, B = – 1.05, t(3) = – 2.34, p < .05. EI did not predict outcomes at follow-up. Results were replicated using the original CIRS measure. Exploratory analyses suggested that the relationship between LI and treatment outcomes was moderated by medication. Involvement in therapy prior to exposures might be an important predictor of posttreatment outcomes in GCBT in a community setting.