Many of Hume's positions have received as much attention as those of any other early modern figure, but his position on knowledge has been surprisingly neglected. In my dissertation, I develop a general interpretation of Hume's position on knowledge. The most striking features of Hume's account, under my interpretation, are that (i) instances of knowledge are immediately present perceptions and (ii) the objects of instances of knowledge are relations between some of their parts. The exegetical and philosophical implications of this account are significant. First, I argue that Hume runs afoul of the widespread contemporary dogma that knowledge entails belief, both in cases involving sense perception and in cases involving abstract philosophical reasoning. Second, I argue that knowledge infallibilisms like Hume's—views that maintain that a knower could not err with respect to what she knows—are compatible with the negation of external world skepticism, contrary to the consensus in the field. Some strains of direct realism provide ample space for the infallibilist to deny this skeptical conclusion. Third, I argue that some of Hume's "demonstrations" do not, in fact, generate knowledge. (Demonstrations are, among other things, sound arguments with necessary premises.) Since Hume is widely interpreted to hold that these demonstrations have literally unbelievable conclusions, this means that Hume must maintain that we can demonstrate claims that we cannot believe.