Examined the role of the working alliance in the treatment of delinquent boys in community-based residential programs, clarifying the relation between therapeutic process and behavioral change. Horvath and Greenberg's (1989) Working Alliance Inventory was used to assess the therapeutic alliance between youth and staff after 3 weeks in treatment and again after 3 months. Achenbach's (1991) Child Behavior Checklist (CBCL; Youth Self-Report [YSR] and Teacher Report Form [TRF]) and recidivism scores were used to assess treatment progress and outcome. Results indicated that a positive working alliance assessed after 3 months in treatment related to positive psychological changes and predicted lower rates of recidivism. Unexpectedly, a positive working alliance assessed early in treatment was associated with negative outcomes (increased internalizing and externalizing symptoms and higher rates of recidivism). This finding suggests that for some delinquent youth initially optimistic assessments may be prognostic of slow progress or treatment failure.