More than a decade has passed since the National Research Council described the common elements of effective educational programs for young children with autism. Since that time, few studies have attempted to understand the mechanisms of change and factors affecting the effectiveness of research supported interventions implemented in community settings. Using Dunst's (2013) framework of implementation science, we examined the relationships between the fidelity of an implementation practice (i.e., a parent-teacher consultation called the Collaborative Model for Promoting Competence and Success; COMPASS), the fidelity of the intervention practice (i.e., teachers' adherence to the intervention plans generated as a result of COMPASS), and child goal attainment outcomes using data from a randomized controlled trial. Results confirmed the predicted relationships between implementation fidelity, intervention practice fidelity, and child outcomes. Specifically, we replicated findings that two hypothesized mechanisms of change, individual education program (IEP) quality and teacher adherence, positively affected intervention practices directly and child outcomes indirectly.