Theses about the nature of logic and science form the core of the various positions attributable to the logical positivists, and logical conventionalism is a particularly interesting case. In this paper I show that Carnap’s brand of logical positivism is committed to a circular logical conventionalism, and offer an explanation of why this is unproblematic. I first explain the motivations for and basic form of Carnapian conventionalism (§1) before introducing Quine’s ‘Tortoise Problem’ (§2), widely considered to refute Carnap’s position. Some have supposed that one can avoid the problem by jettisoning commitment to truth-by-convention: I discuss putative ways to avoid the apparent circularity (§§3-5), showing that one cannot do so without undermining the basic doctrine. Finally I show why commitment to circularity is acceptable (§6): on a more sophisticated account of conventions, circularity does not destroy Carnapian conventionalism.