Mentoring is considered an evidence-based practice for violence prevention. This study presents a partial replication of the "Take Charge!" program implemented in partnership with Big Brothers Big Sisters of America (BBBS). One hundred and eighty-eight early adolescents (M age = 12.87; 61.17% male) who were treated for peer-related assault injury in two urban mid-Atlantic emergency departments were randomly assigned to receive a mentor from two BBBS affiliates. Mentors and organization staff were trained in the "Take Charge!" violence prevention curriculum, which had previously shown evidence of efficacy. Intent-to-treat analyses showed statistically significant improvements in conflict avoidance self-efficacy for the intervention group at 9 months and reductions in fighting at 21 months, but an increase in parental report of aggression at 9 months. Complier average causal effect models revealed evidence of an additional effect for reduced problem behavior at 21 months for intervention adolescents who received a mentor. No effects were found for youth-reported aggression, retaliatory attitudes, deviance acceptance, or commitment to learning. Sensitivity analyses suggested increased aggressive behavior for adolescents in the intervention group who did not receive a mentor (i.e., non-compliers). These findings extend the evidence-base for "Take Charge!" as a violence prevention curriculum for youth already engaged in violence to "real-world" implementation settings. However, they also suggest that challenges associated with providing youth with mentors can be consequential and that additional supports may be needed for these youth/parents.