In this work, I argue for the possibility of epistemic akrasia. An individual S is epistemically akratic if the following conditions hold: (1) S knowingly believes that P though she judges that it is epistemically wrong to do so and (2) Having these mental states displays a failure of rationality that is analogous to classic akrasia. I propose two different types of epistemic akrasia involving different kinds of evidence on which the subject bases her evaluation of her akratic belief. I examine three objections to their possibility. I suggest that the key to defending the possibility of epistemic akrasia is to explain condition (2). I finally argue that epistemic akrasia is possible, and that it represents a failure of mental agency.