The effectiveness of mental health care services is severely limited when young people run away from residential treatment. This study describes the development of a decision support model for predicting discharge due to running away on the basis of individual characteristics. Subjects include 667 wards of a large Midwestern State between the ages of 7 and 20 who were placed in residential treatment and discharged during 2007. A model combining eight predictorsnamely School Attendance, History of Running/Runaway Ideation, and linear as well as quadratic terms for Age, Substance Abuse and Delinquencydemonstrated good internal validity and moderate predictive power for discharge due to elopement. Results of this study can support mental health care workers in identifying potential recipients of interventions that reduce the likelihood of premature exits from residential care among young people. (Published Abstract)