The similarity view of scientific representation has recently been subjected to strong criticism. Much of this criticism has been directed against a ‘naïve’ similarity account, which tries to explain representation solely in terms of similarity between scientific models and the world. This paper examines the more sophisticated account offered by the similarity view’s leading proponent, Ronald Giere. In contrast to the naïve account, Giere’s account appeals to the role played by the scientists using a scientific model. A similar move is often made by defenders of resemblance theories of depiction, who invoke the role played by the artist, or by the viewers of a painting. In this paper I look to debates over depiction to assess the difficulties facing those who wish to defend the similarity view of scientific representation. I then turn to examine Giere’s account. Ultimately, I argue, this account is unsuccessful: while appealing to the role of scientists offers a promising way to defend the similarity view, Giere’s own account does not capture what it is that scientists do when they use a model to represent the world.