A fundamental debate in the philosophy of science is whether our central concepts are true or only useful instruments to help predict and manipulate the world. The first position is termed ‘realism’ and the second ‘instrumentalism’. Strong support for the instrumentalist position comes from the ‘pessimistic induction’ (PI) argument. Given that many key scientific concepts once considered true (e.g. humors, ether, epicycles, phlogiston) are now considered false, how, the argument goes, can we assert that our current concepts are true? The PI argument applies strongly to psychiatric diagnoses. Given our long history of abandoned diagnoses, arguments that we have finally ‘gotten it right’ and developed definitive psychiatric categories that correspond to observer-independent reality are difficult to defend. For our current diagnostic categories, we should settle for a less ambitious vision of truth. For this, the coherence theory, which postulates that something is true when it fits well with the other things we confidently know about the world, can serve us well. Using the coherence theory, a diagnosis is real to the extent that it is well integrated into our accumulating scientific data base. Furthermore, the coherence theory establishes a framework for us to evaluate our diagnostic categories and can provide a set of criteria, closely related to our concept of validators, for deciding when they are getting better. Finally, we need be much less skeptical about the truth status of the aggregate concept of psychiatric illness than we are regarding the specific categories in our current nosology.