We offer a systematic evaluation of the literature on the memorability of supernatural concepts (e.g., gods, ghosts, souls), itself part of a growing body of work in the newly-emerging cognitive science of religion (Barrett, 2000). Specifically, we focus on Boyer’s (1994, 2000, 2001) pioneering Minimally Counterintuitive (MCI) hypothesis according to which supernatural concepts tap a cognitively privilegedmemory-enhancing mechanism linked to violations of default intuitive inferences. Our assessment reveals that the literature on the MCI hypothesis is mired in empirical contradictions and methodological shortcomings which makes it difficult to assess the validity of competing theoretical models, including the MCI hypothesis itself. In light of this fractured picture, we propose a novel and independently motivated account of the memorability of supernatural concepts. This new account is flexible enough to make sense of the heterogenous set of empirical findings in the literature and precise enough to make clear empirical predictions that differ from those of other accounts, including the MCI hypothesis. We conclude with a setof theoretical and methodological prescriptions designed to guide future research on the memorability of supernatural concepts.