We have focused on several features somewhat unique to OA which require special attention relative to clinical trials. These include disease classification, for which the schema in Table 1 was developed, patient selection and admission criteria, and the choice of appropriate clinical outcome parameters. We suggest the following: That future trials be limited to patients with idiopathic OA with specified joint groups involved. That pain, assessed with a visual analogue scale, and both the patient's and investigator's global assessment be the primary clinical outcome variables to be analysed. That concomitant analgesic agents not be used during anti-inflammatory drug trials to prevent confounding pain relief as an outcome variable. That future, long-term studies use validated health status measures to assess functional outcomes.