The structuralist approach represents the relation between a model and physical system as a relation between two mathematical structures. However, since a physical system is prima facienot a mathematical structure, the structuralist approach seemingly fails to represent the fact that science is about concrete, physical reality. In this paper, I take up this problem of lost reality and suggest how it may be solved in a purely structuralist fashion. I start by briefly introducing both the structuralist approach and the problem of lost reality and discussing the various (non-structuralist) solutions that have been proposed in the literature. Following this, I decompose the problem into the ontological mismatch and specification problems. In response to the former, I present a metascientific dissolution argument, according to which the difference in kind between mathematical structures and physical systems poses no deep obstacle to the structuralist approach, and consider some upshots of this argument for our views on representation. By way of conclusion, I argue that the metascientific dissolution argument paves the way for a solution to the specification problem as well.