The paper considers Paul Natorp's Kantian reading of Plato's theory of ideas, as developed in his monumental work, Platos Ideenlehre, eine Einführung in den Idealismus (1903, 1921). Central to Natrop's reading are, I argue, the following two claims: (1) Plato's ideas are laws, not things; and (2) Plato's theory of ideas in the first instance a theory about the possibility and nature of thought - in particular cognitive and indeed scientific or explanatory thought - and only as a consequence is it a theory about the nature of reality. Natrop thus argues that Plato's theory of ideas is at its heart a transcendental theory, and that Plato's metaphysics is built on this basis. The paper considers these claims - and their textual basis in Plato - in some detail, and attempts an initial evaluation of their plausibility as a reading of Plato. I am on the whole sympathetic to Natorp's reading, though a proper assessment goes beyond the present paper. The wider interest of this idealist or anti-realist reading of Plato ought to be obvious, especially in view of the commonly accepted assumption these days that both Plato and Aristotle, and indeed the Greeks in general, took realism entirely for granted (see e.g. M. Burnyeat). Natorp argues that this is true of Aristotle, but quite untrue of Plato. But he is quite clear that the idealism he ascribes to Plato is not Berkeleyan or metaphysical idealism, but a certain kind of transcendental or epistemological idealism. Natorp, however, is no uncritical follower of Kant, and the version of trascendental idealism that he ascribes to Plato is, I argue, very different from Kant's.