John Perry has devoted most of his recent work to the development of what he calls a reflexive–referential theory of utterance content. The theory has two main motivations. The primary purpose of the theory was to offer a new solution in the old controversy between referentialists and descriptivists. The second application of the theory, emphasized in the recent collaboration with Kepa Korta, is to yield a satisfactory compromise in the contemporary debate between minimalists and contextualists. The key idea is that, by acknowledging that any utterance conveys not a single content but a plurality of contents, we can do justice to the intuitions and arguments put forward by all descriptivists, referentialists, minimalists and contextualists at once. The theory recognizes two different kinds of contents, referential and reflexive. The referential content of an utterance u containing a proper name or an indexical expression involves an individual, as the referentialist urges; and, as the contextualist claims, usually that referential content plays the role of what is said by u. In addition to this referential content, u also conveys a variety of other contents: reflexive contents. Each of these corresponds to a different content about u itself, given different facts about u. Perry treats the various reflexive contents of u as descriptive contents, and argues, on behalf of the descriptivist, that some of these capture the cognitive value of utterances like u containing referential terms. Finally, and to pacify the minimalist, one of the reflexive contents of u corresponds to a minimal content, determined independently of context, in virtue of linguistic conventions alone. This paper offers a critical survey of the reflexive–referential theory, highlights its merits, and also points to some of its potential weaknesses.