This article offers a conceptual analysis of collusion, the often overlooked relative of plagiarism in debates on academic integrity. Considered as an inherently social phenomenon, we present the results of a systematic effort to understand the anatomy of collusion. The term's meanings and associated governance practices are compared for contexts outside higher education. These are considered alongside a thematic analysis of publicly available UK university academic integrity documentation that specifies for students what counts as collusion. We indicate how current guideline practice can (1) "appear incomplete" by concentrating on classroom peers, (2) create "blurred boundaries" around useful collaboration, peer review and dishonest practice and (3) be so "unrealistic" as to have unwelcome, unintended consequences for students and staff. Taking an ecological perspective on the conditions of collusion emphasises how these guidelines--by seeking to constrain social interactions around assignment work--may create an uncomfortable incoherence between their prescriptions and well-established patterns of study.