For the purpose of epidemiologic studies in osteoarthritis (OA) an agreed criteria set is required that is both reliable and reproducible. A standardized method of radiologic classification will promote better reporting of epidemiologic data and consistency between centers. The radiographic scoring system of Kellgren and Lawrence has been used in OA research for over 30 years. Its deficiencies include the lack of clarity in interpretation and its reliance on the presence of the osteophyte. Recent advances suggest other pathological evidence of OA on radiographs should be measured when assessing OA. There is growing consensus that while a global score is still useful, the separate features should be recorded in a semiquantitative fashion using standardized atlases allowing different emphasis of particular features at different sites.