Evidence-based assessment (EBA) involves using empirically validated psychological assessment measures to inform case conceptualization and treatment planning. EBA can improve clinical care by providing information relevant for treatment targets and expectations and can clarify how specific client characteristics may be associated with treatment outcomes. Research on EBA practices in relation to particularly challenging clinical problems (e.g., borderline personality disorder) can be especially useful for practitioners. The current study examines Personality Assessment Inventory-Adolescent (PAI-A) predictors of treatment completion, therapeutic alliance, and symptom change in a sample of 44 adolescent boys (Mage = 16.70, SD = 0.56; 75.0% White) in a residential boot camp program for at-risk youths. Participants completed Dialectical Behavior Therapy for Adolescents-informed skills training groups. PAI-A scales related to somatic complaints, depression, antisocial traits, aggression, borderline traits, problematic alcohol use, and interpersonal dominance, as well as an index measuring overall clinical severity, distinguished adolescents who completed the intervention from those who discontinued the intervention. Adolescents with more borderline personality features reported more modest improvements in emotion dysregulation. Additionally, interpersonal warmth was positively related to therapeutic alliance suggesting that client interpersonal style is an important contributor to client-rated therapeutic alliance.