Description:This article has three aims. First, it attempts to identify the exact problem religion poses to those developing an account of the evolution of human social life, and of the cognitive capacities that sustain that life. Having done that, the article aims to specify the criteria that an explanation of the evolution of religion should meet. Second, it presents a critique of one popular account: the idea that religion emerges as a result of cognitive biases that favor agentive thinking, that favor minimally counterintuitive narratives, amplified and entrenched by some form of cultural evolution. Third, it sketches a positive account in which various forms of communal practice and quasi-religious activity precede the emergence of religious belief. The machinery of the cultural transmission of religion is built before religion transitions into (in part) an ideological system. I call this an ethnographic model of the evolution of religion.