Some critics argue that the type of psychiatric diagnosis found in the DSM and ICD are superfluous and should be abandoned. These are known as categorical polythetic psychiatric diagnoses. To receive a categorical polythetic psychiatric diagnosis an individual need only exhibit some, rather than all, of the symptoms on the diagnostic criteria. Consequently, categorical polythetic psychiatric diagnoses only associate an individual with a range of symptoms rather than specify which symptoms they have. Drawing upon Ronald Giere’s account of scientific models, I portray categorical polythetic psychiatric diagnoses as abstract models which guide the building of less abstract models. These models can specify which symptoms a particular individual has. Additionally, categorical polythetic psychiatric diagnoses can guide investigation of symptoms towards difficult to spot symptoms, guide investigation towards changing symptoms and guide investigation towards how symptoms manifest. These important roles mean categorical polythetic psychiatric diagnoses should not be abandoned.