Over the last decade or so philosophers have become increasingly dissatisfied with the dominant focus on the notion of knowledge in contemporary epistemology. Understanding, it has been suggested, is an important intellectual goal, often valuable over and above knowing the facts. But there is much disagreement over how exactly this suggestion is to be spelt out. Does understanding consist in a specific form of knowledge? Or is it a cognitive state intrinsically distinct from, and irreducible to, knowledge? And if the latter, how does understanding differ from knowledge? In this paper, I sketch an answer to these questions which I attribute to Kant. By contrast with contemporary accounts, Kant’s primary epistemological concern lies not with knowledge but with cognition and its intuitive and conceptual conditions. What contemporary epistemologists call ‘understanding’ is, for Kant, a type of cognition. Knowledge, by contrast, does not present a form of cognition and, hence, differs in important respects from understanding. This conclusion, however, does not imply a multiplicity of potentially conflicting cognitive goals. The important insight of Kant’s account, I argue, is that the search for understanding is an indispensable means in the search for knowledge.