Several authors have recently defended the thesis that whether someone knows something can depend on actual or perceived practical factors of her situation, in addition to whether she believes it, the reasons for which she believes it, the available evidence, the truth-conduciveness of her environment, whether her faculties are reliable, and such other mundane factors on which knowledge is ordinarily allowed to depend. Call this the pragmatic encroachment thesis.1 According to John Hawthorne [2004] the pragmatic encroachment thesis is required in order to solve certain puzzles deriving from thinking about lottery cases. According to Jason Stanley [2005] it is directly motivated by intuitions about cases. And according to Jeremy Fantl and Matthew McGrath [2010], it is the best way to make sense of the fallibility of knowledge. This paper also explores the pragmatic encroachment thesis, and also as a sympathetic voice. But rather than arguing that the pragmatic encroachment thesis is true, my primary task will be to try to make sense of how it could be true. For the most obvious problem with the pragmatic encroachment thesis is that it doesn’t seem like the kind of thing that could be true. We’re all brought up in epistemology on the foil of Pascal’s Wager, which shows how it might be advisable or beneficial to have some belief, independently of whether it is true. But Pascalian considerations, we observe, can’t ground knowledge. Moreover, we observe, if there is any sense of ‘rational’ in which you don’t know something unless it is rational for you to believe it, Pascalian considerations can’t affect this kind of rationality – usually called epistemic rationality – either. If Pascalian considerations are our paradigm of practical factors, then it is puzzling in the extreme how practical factors could affect knowledge. It is this challenge that I aim to address in this paper. What I aim to do, is to show how practical factors could play a role in defeating knowledge by defeating epistemic rationality – the very kind of rationality that is entailed by knowledge, and in which Pascalian considerations do not play any role...