Representation theorems—either those that take us from a ranking on propositions to a unique probability function, or those that allow for the representation of a total preference state by a probability and utility pair—have very often been taken to be of some kind of importance for the foundations of decision theory. One tradition, beginning with Frank Ramsey, through de Finetti, Savage, and others, holds that representation theorems are useful in descriptively characterising what it is to have such-and-such degrees of belief; a closely related tradition seeks to use these theorems in the service of arguments for probabilism, a normative thesis about our degrees of belief. Meacham & Weisberg (AJP, 2011) argue that these theorems can play no such foundational role, and that their importance to decision theory is minimal at best. This paper is a defence of the importance of representation theorems. I argue that while there are very good reasons to reject many simplistic positions in the vicinity, there are views that fall within the Ramseyean tradition not touched by the arguments of Meacham & Weisberg. I also provide some reasons to doubt the use of these theorems as premises in arguments designed to support probabilism.