Focus constructions have often been seen as exceptions to Condition C. In the sentences Only JOHN likes John and John only likes JOHN, the two occurrences of "John" may co-refer given the proper discourse context, even though one c-commands the other. I argue that focus constructions are not exceptions to Condition C, but follow straightforwardly from standard accounts of binding and focus. I adopt the binding theory of Safir (2004), a competition-based theory, which roughly states that Condition C is a result of the fact that a bound R-expression can normally be replaced with an anaphor or pronoun without affecting the interpretation, and anaphors or pronouns are preferred over R-expressions when they are available. I argue that focus constructions with bound R-expressions are grammatical because replacing the bound R-expression with an anaphor would in fact change the interpretation, taking into account Rooth's (1992) proposed second component of meaning in focus constructions, the so-called focus semantic value. For cases like John only likes JOHN, I argue that they are grammatical for pragmatic reasons. Safir's theory allows for an override of the binding conditions for pragmatic reasons in certain cases, and I argue that this is such a case. As is well known, focus constructions are subject to a parallelism condition—one of the propositions which the assertion is being contrasted with must be present (at least implicitly) in the discourse. I argue that John only likes JOHN is felicitous because it satisfies a structural parallelism condition which John only likes HIMSELF does not.