There are advantages to be gained from a knowledge of the patient casemix that psychiatric hospitals, and community services, are likely to encounter. In particular the efficient allocation of resources would be better achieved if individuals could be ‘placed’ into groups. These groups would need to be clinically distinct and contain similar members. Previous methods of defining casemix have focused upon clinical diagnosis. However, the effectiveness of diagnosis in predicting resource utilisation is limited. A range of other factors could be used, including comorbidity, severity of illness, and previous hospital contact.