Three experiments investigate whether and why people accept explanations for symptoms that appeal to mental disorders, such as: “She experiences delusions because she has schizophrenia.” Such explanations are potentially puzzling, as mental disorder diagnoses are made on the basis of symptoms, and the DSM implicitly rejects a commitment to some common, underlying cause. Do laypeople nonetheless conceptualize mental disorder classifications in causal terms? Or is this an instance of non-causal explanation? Experiment 1 shows that such explanations are indeed found explanatory. Experiment 2 presents participants with novel disorders that are stipulated to involve or not involve an underlying cause across symptoms and people. Disorder classifications are found more explanatory when a causal basis is stipulated, or when participants infer that one is present (even after it's denied in the text). Finally, Experiment 3 finds that merely having a principled, but non-causal, basis for defining symptom clusters is insufficient to reach the explanatory potential of categories with a stipulated common cause. We discuss the implications for accounts of explanation and for psychiatry.