This dissertation is a defense of moderate invariantism, the traditional epistemological position combining the following three theses: invariantism, according to which the word ‘know’ expresses the same content in every context of use; intellectualism, according to which whether one knows a certain proposition does not depend on one’s practical interests; and antiskepticism, according to which we really do know much of what we ordinarily take ourselves to know. Moderate invariantism needs defending because of seemingly powerful arguments for contextualism, the view (...) that, like ‘I’ and ‘now’, ‘know’ expresses different contents in different contexts. It has been argued that only contextualism can properly reply to skeptical arguments, and that only contextualism can account for our tendency to go from judging that a knowledge claim is true to judging that it is false in response to shifts in the context of use. Moderate invariantist replies to these arguments have largely failed. I propose new replies on behalf of moderate invariantism, while critiquing earlier, less successful attempts. Chapter one is introductory. In chapters two and three, I examine contextualist replies to skeptical arguments arising from radical skeptical hypotheses (such as the possibility that one is a brain in a vat) and from considerations involving lotteries. I argue that if contextualism can adequately respond to these arguments, then there are equally effective replies available to the moderate invariantist. In the next two chapters, I examine Stewart Cohen’s “Airport Case,” which elicits intuitions about knowledge claims that supposedly only contextualism can accommodate. In chapter four, I argue that none of the invariantist replies to the case that depend on denying the intuitions succeed; in chapter five, I accept the intuitions, but argue that contextualism does not follow from them. Finally, in chapter six, I evaluate two interesting invariantist critiques of contextualism; according to the first, contextualism collapses into the radical position that every English expression is context-sensitive; according to the second, ‘know’ fails a test for context-sensitivity involving indirect speech reports. (shrink)