Oxford Studies in Epistemology. forthcomingAccording to reliabilism, whether a belief is justified, or amounts to knowledge, is a matter of reliability. We consider familiar arguments that reliabilism is incompatible with defeat, and show how reliabilists can respond to them by understanding reliability in a more flexible way—roughly, whether a belief is reliable is a matter of how likely beliefs like it are to be true. The most straight-forward way of using this flexibility over-generates justification in uplift cases, cases in which you learn but ignore excellent evidence for an otherwise unjustified belief. We argue that this problem is much harder, and calls for a structural revision of reliabilism. We propose our favorite such revision, which helps the reliabilist draw the distinction between a belief being propositionally and doxastically justified.