Abstract In this paper I discuss Polycarp Ikuenobe's view that it is rational to believe, in an African context, in the existence of witches and witchcraft. First, I attempt to show that it is not possible to prove empirically that witches and witchcraft are real, as Ikuenobe assumes. I argue that even though witches and witchcraft are part of the social reality in which many Africans live, they do not have the same ontological status as theoretical entities in scientific research. (...) Second, I try to show that Ikuenobe's attempt to demonstrate that the belief in witches and witchcraft has a rational foundation is not convincing. Admittedly, Africans, who live in magic-ridden cultures, have reasons that locally justify their belief in witches and witchcraft. However, when the justification offered for this belief is assessed by external standards, employed within scientific discourse, it turns out to be insufficient. (shrink)