Many philosophers have argued or taken for granted that Frege's puzzle has little or nothing to do with identity statements. I show that this is wrong, arguing that the puzzle can only be motivated relative to a thinker's beliefs about the identity or distinctness of the relevant object. The result is important, as it suggests that the puzzle can be solved, not by a semantic theory of names or referring expressions as such, but simply by a theory of identity statements. To show this, I sketch a framework for developing solutions of this sort. I also consider how this result could be implemented by two in uential solutions to Frege's puzzle, Perry's referential-re exivism and Fine's semantic relationism.