Approximately a dozen internal research studies into student learning and program effectiveness were conducted during ArtsSmarts' first eight years. In the spring of 2006, we compiled the results of those studies, along with a like number of reports by outside researchers, to create a synthesis of possible directions for future work. Although we used a small sample of available outside studies, it was immediately and glaringly evident that the arts and educational communities are hungering for research that will "help us understand what the arts learning experience is for children, and what characteristics of that experience are likely to travel across domains of learning" (Deasy, 2002:99). It was equally evident to all ArtsSmarts partners that, while future ArtsSmarts research could be taken in any number of directions, it made the most sense to identify and build from ArtsSmarts' own strengths and successes. We also felt the need to align the research direction and the methods of data collection with our intended audiences.Different groups would find different aspects of ArtsSmarts compelling, and distinctly different types of data would be required for each. Partners identified educators (teachers, administrators, and senior Board office personnel) as the audience they most wanted to reach.With that in mind, the decision was made to develop a theory of learning that would serve the dual purposes of explaining ArtsSmarts' impact in Canadian classrooms and framing the research work of the next few years. We felt that establishing an ArtsSmarts theory of learning would help to answer the question, "If ArtsSmarts didn't exist, what would be lost?" Further, a theory of learning would assist teachers, artists and partners in identifying key, essential components of the ArtsSmarts experience, and would also prevent ArtsSmarts from being viewed as a pleasant but unnecessary add-on to classroom activity. The paper that follows develops an ArtsSmarts theory of learning centred on the concept of student engagement.