<p>Some philosophers have suggested that an analogue of Moore’s paradox can be formulated for the case of inference, and that this reformulated paradox would have to be explained by any acceptable account of the nature of inference. In this paper, I critically discuss the current formulations of Inferential Moore’s Paradox (IMP) and suggest some improvements. I then provide an argument for the thesis that inference is conceptually governed by a normative standard of correctness and offer what I think is the best formulation of this standard. Finally, I show how inference’s conceptual standard of correctness can be deployed to explain IMP.</p>