By considering the semantic footings of the so-called antinomies of pure reason, this article contributes to the debate about whether Kant was committed to semantic realism or anti-realism. That is, whether verification-transcendent judgements are truth-apt (realism) or not (anti-realism). Against the (empiricist) semantic principle that Strawson, and others, have ascribed to Kant as the “principle of significance,” the bedrock of my article is what I call Kant’s Real Principle of Significance: an extension-based and normative principle stating that a judgement can (...) have no “significance” or “objective validity” (truth-value) without a universally recognizable norm for verifying it. This principle entails semantic anti-realism. I argue that we can extract the principle from the antinomy chapter of KrV, since in there Kant concludes that judgements of the form “the world as such is x” are without “significance” (lack a truth-value) in virtue of being unverifiable as a matter of principle. I propose that Kant’s reference to some of the antinomical judgements as “false” is not incompatible with this anti-realist reading because he operates with two distinct world-concepts: an illegitimate transcendental realist one and a legitimate transcendental idealist one. In contrast to most anti-realist Kant-interpretations, it is furthermore argued that any satisfactory anti-realist construal of Kant’s view must be compatible with his assertion that the thesis in the third antinomy about freedom “can be true.” That requires a thicker conception of “significance” or “objective validity” than what is often ascribed to Kant, which is encapsulated by the Real Principle of Significance. (shrink)