Diegesis is central to film criticism, yet, at least in film studies, it is a conceptual mess. Its definition in the discipline contradicts its original meaning and relies upon poorly theorised models of story worlds, producing an ambiguous boundary between diegetic and non-diegetic properties. Determinations of diegetic boundaries can be so contestable they lack critical utility. Plato and Aristotle define diegesis as narration, which stands in relation to mimesis, or imitation. Yet in the 1950s, French film theorists effectively conflate these terms, using diegesis to refer to a film’s narration and its content. Subsequently, the definition of diegesis shifted further, to distinguish representations that are part of the story world from those that are not. This delineation was then erected on two problematic theories: (1) idealist conceptions of story worlds and (2) illusion theories of filmic images and sounds. This framework undermines the utility of diegesis to evaluate the aesthetic qualities of cinematographic works. I argue that the objectives behind the modern definition of diegesis can be rescued by reconsidering the classical coupling of diegesis and mimesis, and attaching that to a realist account of fiction and fictive iconic reference.