As pressure increases to reduce spending on residential treatment and shift scarce mental health and child-welfare dollars to community-based forms of care, interest has risen in identifying factors that effect length of stay (LOS) in residential treatment. Existing data from a single residential treatment center located in the northeastern United States were used to examine the affect of an array of predisposing, enabling, illness, and current treatment factors on the probability of discharge from residential treatment for 165 children with serious emotional disturbance. Results of the discrete-time survival analysis indicated that while the main effects of number of prior psychiatric hospitalizations, externalizing and internalizing scores on the Devereux Scales of Mental Disorders (DSMD), and rate of physical restraint were significant, only rate of physical restraint retained significance when these variables were added together into a discrete-time hazard model. Results suggest that while predisposing, enabling, and illness factors may be important in determining who receives residential treatment, those factors related to the treatment at-hand are important determinates of how long a child will actually spend in care once admitted. Implications for social work policy and practice as well suggestions for future research are also discussed.