In this article I aim to clarify the nature of Kant’s transformation of rationalist metaphysics into a science by focusing on his conception of transcendental reflection. The aim of Kant’s Critique of Pure Reason, it is argued, consists primarily in liberating the productive strand of former general metaphysics – its reflection on the a priori elements of all knowledge – from the uncritical application of these elements to all things (within general metaphysics itself) and to things that can only be thought (in special metaphysics). After considering Kant’s conception of metaphysics and his various uses of the term ‘transcendental’ I closely examine his account of logical and transcendental reflection in the section entitled ‘On the Amphiboly of the Concepts of Reflection’. Whereas commentators generally attribute the activity called transcendental reflection to Kant alone, I contend, first, that Kant regarded philosophy as such to rely on a mode of transcendental reflection and, second, that the critical mode of transcendental reflection enacted in the Critique itself yields insight into the reason why our a priori knowledge is limited to the realm of possible objects. This is illustrated by outlining the difference between Kant’s and Leibniz’ employment of the concepts of reflection.