DescriptionUnder what conditions do we have inferential knowledge? I propose and defend the following principle: S knows that p via inference only if S knows all the premises essentially involved in her inference in support of p - “KFK” for short. Even though KFK is at least tacitly endorsed by many figures in the history of philosophy, from Aristotle through Descartes, and Kant to Bertrand Russell – and, more recently, by David Armstrong – KFK has fallen into disfavor among epistemologists over the past fifty years. In response to Edmund Gettier’s legendary paper, many have proposed views according to which one’s reasoning is a source of knowledge even if one fails to know some or all premises essentially involved in one's reasoning, while others have given up offering a theory of inferential knowledge and have focused on reasoning as a source of justified belief instead. Unfortunately, these accounts that deal with inferential knowledge are problematic; they cannot, for example, fully explain our common practice of evaluating negatively inferences with unknown premises. They also seem to rely on an overly narrow understanding of the so-called “Gettier Problem”. This dissertation aims at updating the approach to reasoning that was popular before Gettier by building on the framework of Timothy Williamson's “knowledge-first” epistemology. The aim is to carefully elaborate the view that reasoning yields knowledge only if all the premises essentially involved are also known.