A total of 94 children (80 percent) participated in therapy. Of these, the attrition rate was 20 percent. Demographic variables--the child's age, race, or gender--were not predictive of leaving treatment. Whether or not the treatment was mandated was not predictive of attrition. Findings strongly indicate that the caregivers who, at intake, reported high levels of distress related to the caregiving role were at the highest risk for not returning to treatment beyond the first assessment session. This suggests that this group of caregivers might benefit from additional outreach and efforts to engage them in the treatment process from the outset. A caregiver report of social support was not predictive of attrition from treatment. Caregivers who remained in treatment reported significantly higher levels of problem behavior. Families who received multimodal treatment and care were more likely to complete treatment than those assigned to individual-only or family-only treatment. Making a child abuse report during treatment was unrelated to attrition. The sample included children ages 4-17 who had been referred for treatment of child abuse and/or exposure to domestic violence at a pro bono outpatient pediatric mental health clinic at the University of Maryland in Baltimore. Data were obtained from computerized comprehensive assessment protocols completed by the child and his/her primary caregiver, referral forms, progress notes, and discharge summaries. Data were entered into a statistical software program for analyses. 3 tables and 42 references