It is natural to think that law ought not to be vague. After all, law is supposed to guide conduct, and vague law seems poorly suited to do that. Contrary to this common impression, however, a number of authors have argued that vagueness in the law is sometimes a good thing, because it is a means to achieving certain valuable legislative ends. In this article, I argue that many authors—including Timothy Endicott and Jeremy Waldron—wrongly associate vagueness with instrumental roles that are really played by a closely related semantic phenomenon.