For over 30 years, investigators have used the simple but nonvalidated classification criteria suggested by Moll and Wright. Several authors have suggested modifications to these, but most remain invalid or require human leukocyte antigen analysis. Now, a worldwide initiative has developed new criteria which include both clinical and radiological features. These will require further study before they are fully adopted for future studies, although improved performance should result in less variation between study cohorts. The recurring question of disease heterogeneity continues to occupy researchers in this field. Despite recent pleas to abandon the original five subgroups, a case can be made for retaining at least the two subgroups of peripheral and axial disease and, possibly, splitting the peripheral disease into oligo- and polyarthritis.