Donald Davidson's work on the theory of meaning has been enormously influential since the publication of “Truth and Meaning” in 1967. His central proposal was that an understanding of what it is for “words to mean what they do” (Davidson 1984b, p. viii) can be pursued by way of constructing and confirming for a speaker an axiomatic truth theory, modeled on a Tarski-style axiomatic truth definition, for his language. In this chapter, we first discuss the background of Davidson's famous suggestion, initially introduced in “Truth and Meaning.” We begin with his arguments for the importance of attending to the compositionality of natural languages in §1, then turn in §2 to his criticisms of traditional approaches to the theory of meaning. In §3, we discuss Davidson's introduction of a truth theory as the vehicle for a compositional meaning theory; in §4, we explore some interpretive issues that arise about his intentions, specifically the question of whether Davidson intended to replace the traditional project (providing an account of meaning) with a more tractable one (providing an account of truth conditions), or whether he intended to pursue the traditional project by novel means. We argue that, though Davidson has been widely misunderstood, his intention is clearly the latter, and, specifically, that his goal has always been to give an account of what illuminating constraints a truth theory can meet that would suffice for it to be used to understand any potential utterance of an object language sentence.