The notion of exemplification is essential for Goodman’s theory of symbols. But Goodman’s account of exemplification has been criticized as unclear and inadequate. He points out two conditions for an object x exemplifying a label y: (C1) y denotes x and (C2) x refers to y. While (C1) is uncontroversial, (C2) raises the question of how “refers to” should be interpreted. This problem is intertwined with three further questions that consequently should be discussed together with it. Are the two necessary conditions (C1) and (C2) conjointly sufficient? Do they amount to a definition of “exemplification”? Which notions of Goodman’s theory are basic, and hence undefined? In this paper, we address these questions and defend a reconstruction of the notion of exemplification that interprets “refers to” in (C2) as exemplificational reference and hence treats “exemplification” as a basic notion of Goodman’s theory. Firstly, we argue that even though the notion of exemplification is not defined, it is still sufficiently clear. This ensures its contribution to Goodman’s theory of symbols. Secondly, we show that our account is plausible as an interpretation of Goodman’s and Elgin’s writings, although it implies that some of Goodman’s theorems about self-reference have to be weakened. Thirdly, we argue that it is the only materially adequate reconstruction of Goodman’s notion of exemplification, whereas the alternative definitional accounts fail.