H. P. Grice is justly famous for his work on implicature. He identified a theoretically important phenomenon and provided a theory that is still dominant. Grice's virtual discovery of implicature was a major achievement, a breakthrough in linguistics and philosophy of language. Nevertheless, his theory of implicature, I argue here, is a near-complete failure.

After explaining the concept of implicature and illustrating its theoretical importance, I review Grice's theory. It is distinguished by the claim that conversational implicatures arise from certain principles of cooperative behavior as they apply to conversation. I isolate and analyze four principal elements in Gricean theory: the Theoretical Definition of conversational implicature in terms of the conversational principles; the Calculability Assumption, which holds that conversational implicatures can always be worked out or inferred from the conversational principles; the Generative Assumption, which claims that conversational implicatures exist because of facts involving the conversational principles; and Grice's Razor, which holds that it is more economical to postulate conversational implicatures rather than senses because conversational implicatures can be derived from independently motivated principles. I use the term Gricean Theory broadly to denote not just Grice's own theory, but any principle-based theory with these four or similar elements.

I develop four main lines of argument against Gricean theory. First, the theory generates erroneous predictions as readily as it generates correct ones. Indeed, it is the exception rather than the rule to find the implicatures predicted by Gricean derivations. Second, the conversational principles have insufficient power to enable rigorous derivation of specific implicatures.