To date, many treatment programs for troubled youth have been shown to be effective on the basis of rigorous studies, but the processes responsible for the success of these programs are still unknown. This article presents the study of mechanisms of change as a promising research avenue for optimizing treatment programs based on their capacity to identify why, how, for whom, and in what contexts they work best. The theory-driven approach to program evaluation is proposed as a useful framework to guide the identification of potential mediating and moderating variables in treatment outcomes. Finally, conceptual and methodological clarifications are provided, illustrated by examples based on treatment programs for children with problematic sexual behaviour or for juvenile sex offenders.