Systems of Care (SOC) have demonstrated modest benefits for youth with serious emotional and behavioral disturbances; however, it is uncertain that youth benefit more when they receive treatment in a SOC relative to traditional treatment. Although considerable research has examined the degree to which youth enrolled in a SOC improve over time, no published studies were found that specifically investigated the degree to which improvements were maintained following SOC discharge. The maintenance of treatment gains could further support the use of SOC principles, and variability in the degree to which treatment effects are maintained may encourage greater emphasis on long-term sustainability of treatment effects. In theory, implementation of SOC principles should promote sustained treatment gains (e.g., by building a support system outside formal service provision); however, prior research has documented inconsistent implementation of specific elements of the philosophy that are expected to contribute to the maintenance of treatment gains. As such, the present study sought to examine the degree to which treatment effects were maintained or lost following SOC discharge, and the factors that predicted youths’ post-discharge trajectories. The present study found that youth who were enrolled in the SOC improved in functioning (reduced symptoms) over time (i.e., between the time of enrollment and up to three years following); however, it was necessary to account for trajectory changes occurring post-discharge to provide a more accurate estimation of improvements. Initially, without accounting for discharge, significant improvements over time were not detected because improvements were measured as a constant rate, and a diminished rate of improvement following discharge produced error that masked significant improvements. Examining the trajectory change following discharge suggested that youth improved significantly over time, but that the rate of improvement decreased following discharge. That is, youth did not tend to "relapse" or worsen in functioning following discharge but, rather, often continued improving, albeit at a slower rate than during treatment. Older youth improved more during SOC enrollment than younger youth; while younger youth improved more than older youth following discharge from the SOC. The discrepancies between system-level enrollment data and caregiver-reported services also were used in predicting youth improvement trajectories before and after discharge. Youth who were reportedly not served for at least one 6-month period prior to the date of discharge (as reported by the SOC; i.e., served inconsistently) improved more slowly during SOC enrollment and improved less overall compared to youth who received services consistently throughout enrollment. In addition, youth who reportedly received services following the date of discharge tended to experience greater overall severity of symptomatology over time; however, they improved relatively quickly during SOC enrollment, were discharged sooner, and later exhibited a lower rate of post-discharge improvement compared to youth who did not receive post-discharge services. A number of caregiver-reported ecological variables were also examined in relation to youth trajectories, including the proportion of days youth were treated in out-of-home placements (e.g., residential care), the quality of family interactions, familial risk factors (e.g., homelessness, domestic violence), the degree to which caregivers experienced strain in caring for their child, and the amount of natural support (i.e., not from paid providers) caregivers reportedly received. The predictor variables of focus did not relate significantly to youth trajectories in these analyses. Missing data and the large number of parameters tested may have limited the ability to detect relationships between youth functioning trajectories and predictor variables. The results from the present study have implications for the way in which longer-term improvements in treatment settings are evaluated, as they underscore the need for longitudinal analytic designs to account for trajectory changes at discharge. If such models omit the effect of discharge, they assume improvements are made at a continuous rate over time, which can mask treatment effects or relative advantages of an intervention over time. Furthermore, results suggest that improving the consistency of service receipt during SOC enrollment may be a relatively tangible mechanism for quality improvement in service provision.