<p>There has been a sharp increase in the use of impossible worlds as theoretical tools for solving difficult philosophical problems. Some philosophers, however, warn against their use. For example, Timothy Williamson argues that impossible worlds should not be used in an analysis of conditionals because they do not provide a compositional semantics. In this paper, we set out to resolve some of the potential problems associated with impossible worlds, thereby providing justification for their uses in a variety of contexts and applications. Graham Priest provides two directives for an account of impossible worlds. We argue that Priest’s directives, and, thus, the traditional view on impossibility, overextend the class of impossible worlds. We argue that all worlds, possible or impossible, are describable and that nonsense cannot play a role in a proper description. So, putative worlds that include nonsense are not worlds at all, possible or impossible. The exclusion of nonsense sentences from worlds undermines Priest’s two directives, but by separating nonsense from impossibility, we can resolve many of the worries about the use of impossible worlds.</p>