While the measurement of treatment integrity is important to determine how much, and how well, interventions are delivered in schools, the science of treatment integrity is not well developed in education research. The purpose of this paper is to describe a program of research that has developed treatment integrity measures over the past 10 years to assess teacher delivery of an indicated program targeting reductions in problem behavior in early childhood and elementary school classrooms. Specifically, this paper will highlight the importance of active use of conceptual models to guide treatment integrity measure development, multidimensional assessment of treatment integrity and training procedures for observers, using several studies to illustrate the evolution and refinement of our measurement approach. Recommendations for researchers developing and evaluating interventions in schools are provided, as are recommendations to help the field move toward a more rigorous science of treatment integrity.