The number of people with arthritis in the US was 43 million in 1998, and is projected to reach 60 million by 2020. The burden of arthritis is substantial and is responsible for $65 billion annually in direct medical costs and lost productivity. The most common form of arthritis, osteoarthritis (OA), currently affects over 20 million adults in the US. Because OA has no known cure, the goals of therapy are symptom relief and improvement in patient-reported outcomes. OBJECTIVES: Our objective was to systematically review disease-specific functional health status instruments recommended in consensus guidelines for clinical research in OA (as established by Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT) and OsteoArthritis Research Society International (OARSI)) and to assess their ability to predict economic outcomes. METHODS: Computerized literature searches were conducted in the following databases from 1978-2002: EMBASE, HealthSTAR, MEDLINE, and PSYCHINFO. Manual searches were also conducted. The six disease-specific instruments recommended by the OMERACT/OARSI guidelines were reviewed with regard to their domains, psychometric properties, and usefulness in clinical research: the Health Assessment Questionnaire, Arthritis Impact Measurement Scales, Western Ontario and McMaster Universities Arthritis Index, Indices of Clinical Severity, Australian/Canadian Osteoarthritis Hand Index, and the Algofunctional Index. RESULTS: Substantial variation in the psychometric properties and domains covered by the instruments was found. Physical function and pain were the domains which were most frequently emphasized and which seemed to have the most value in predicting economic burden of disease. Underrepresented areas included the mental, social and emotional aspects of OA. CONCLUSIONS: Additional research should evaluate the usefulness of various instruments in assessing the impact of OA treatments through change in physical and psychological health status over the long term. Further evaluation of instruments' predictability of economic outcomes is needed.