Abstract: Various structural requirements of rationality prohibiting a kind of incoherence within a subject’s mind have been immensely popular. I focus, in particular, on enkratic requirements of rationality, according to which it is irrational, for instance, to believe p, while believing that it is irrational for one to believe p (or believing that one’s evidence does not support p). There is a general worry about how such structural requirements are to be reconciled with what one has epistemic reason to believe. Nevertheless, it has been argued that at least the following is true: if a subject believes as her epistemic reasons require her to believe, she conforms to structural requirements of rationality. However, I argue that this is not the case: there is a tension between widely accepted norms on belief and the enkratic requirements. This gives rise to a resilient paradox. I discuss and reject various solutions, and sketch a way out. I sketch the beginning of an account that rejects the enkratic requirements as genuine requirements of rationality, while nevertheless explaining their appeal. I try to generalize the account to other putative structural requirements.