The authors hypothesized that the experience of maltreatment would interfere with alliance formation, thereby compromising the effectiveness of therapy. Results indicated that maltreatment, multiplicity of maltreatment, and type of perpetrator of maltreatment were all associated with initial alliance difficulties. None of these variables predicted change in the alliance over time; instead, severity of interpersonal problems was the best predictor of alliance development. Maltreatment status was not directly related to treatment outcome; however, maltreated adolescents who failed to develop positive alliances with their therapists tended to show the poorest outcomes. Recognizing the strong relationship between alliance formation and outcome, the authors have included clinical recommendations for therapists who treat maltreated adolescents. Tables, references