A foundational question for the discipline of psychiatry is the nature of psychiatric disorders. What kinds of things are they? In this paper, I review and critique three major relevant theories: realism, pragmatism and constructivism. Realism assumes that the content of science is real and independent of human activities. I distinguish two “flavors” of realism: chemistry‐based, for which the paradigmatic example is elements of the periodic table, and biology‐based, for which the paradigm is species. The latter is a much better fit for psychiatry. Pragmatism articulates a sensible approach to psychiatric disorders just seeking categories that perform well in the world. But it makes no claim about the reality of those disorders. This is problematic, because we have a duty to advocate for our profession and our patients against other physicians who never doubt the reality of the disorders they treat. Constructivism has been associated with anti‐psychiatry activists, but we should admit that social forces play a role in the creation of our diagnoses, as they do in many sciences. However, truly socially constructed psychiatric disorders are rare. I then describe powerful arguments against a realist theory of psychiatric disorders. Because so many prior psychiatric diagnoses have been proposed and then abandoned, can we really claim that our current nosologies have it right? Much of our current nosology arose from a series of historical figures and events which could have gone differently. If we re‐run the tape of history over and over again, the DSM and ICD would not likely have the same categories on every iteration. Therefore, we should argue more confidently for the reality of broader constructs of psychiatric illness rather than our current diagnostic categories, which remain tentative. Finally, instead of thinking that our disorders are true because they correspond to clear entities in the world, we should consider a coherence theory of truth by which disorders become more true when they fit better into what else we know about the world. In our ongoing project to study and justify the nature of psychiatric disorders, we ought to be broadly pragmatic but not lose sight of an underlying commitment, despite the associated difficulties, to the reality of psychiatric illness.