Abstract
On the basis of arguments showing that none of the most influential analyses of Moore's paradox yields a successful resolution of the problem, a new analysis of it is offered. It is argued that, in attempting to render verdicts of either inconsistency or self‐contradiction or self‐refutation, those analyses have all failed to satisfactorily explain why a Moore‐paradoxical proposition is such that it cannot be rationally believed. According to the proposed solution put forward here, a Moore‐paradoxical proposition is one for which the believer can have no non‐overridden evidence. the arguments for this claim make use of some of Peter Klein's views on epistemic defeasibility. It is further suggested that this proposal may have important meta‐epistemological implications.