What Are Symmetries?

I advance a stipulational account of symmetries, according to which symmetries are part of the content of theories. For a theory to have a certain symmetry is for the theory to stipulate that models related by the symmetry represent the same possibility. I show that the stipulational account compares positively with alternatives, including Dasgupta's epistemic account of symmetry, Moller-Nielsen's motivational account, and so-called formal and ontic accounts. In particular, the stipulational account avoids the problems Belot and Dasgupta have raised against formal and ontic accounts of symmetry while retaining many of the advantages of these otherwise-attractive frameworks.