Tarski notoriously argued, from consideration of the Liar paradox, that no consistent language can be semantically closed, and our only option for a consistent theory of truth is a hierarchy of languages. These claims have been subject to much criticism over the years. Many hold them to be prima facie implausible, and some hold that Tarski’s notion of a consistent language is itself a confusion. I, on the other hand, have long advocated for a highly revised form of a Tarskian hierarchy. In this paper, I return to the question of how to formulate claims about semantic closure, and I argue that a properly formulated hierarchical theory of truth is more plausible and natural than it is usually taken to be.