Abstract. Despite the growing prominence of canonical action research (CAR) in the information systems discipline, a paucity of methodological guidance continues to hamper those conducting and evaluating such studies. This article elicits a set of five principles and associated criteria to help assure both the rigor and the relevance of CAR in information systems. The first principle relates to the development of an agreement that facilitates collaboration between the action researcher and the client. The second principle is based upon a cyclical process model for action research that consists of five stages: diagnosis, planning, intervention, evaluation and reflection. Additional principles highlight the critical roles of theory, change through action, and the specification of learning in terms of implications for both research and practice. The five principles are illustrated through the analysis of one recently published CAR study.