I address a longstanding problem in Kant scholarship: how is Kant’s use of the term ‘mechanism’ to be understood? It seems that Kant uses that term in a variety of ways, from a narrow sense (“motion communicated between matter”) to a very wide sense (“any causation that is not noumenal”). I argue that Kant has a unified conception of mechanism, where the wider senses are to be understood in light of a conception of nature according to which all of nature (...) is purely mechanistic in the narrow sense. In contrast to his mechanistic predecessors, Kant holds that this conception of nature is an idea, an imaginary end-point of science that we need for orientation but can only approach asymptotically. (shrink)