To compare the responsiveness of arthritis-specific and generic health outcome measures in relation to changes in Rheumatoid Arthritis (RA) severity and to treatment. Clinical trial patients (N=315) were assessed at baseline and again after two weeks of treatment. Criterion measures of severity change included physician global assessment, duration of morning stiffness, number of tender and swelling joints, functional capacity classification, visual analogue pain scale, and patient global assessment. METHOD: The responsiveness of each disease-specific and generic measure was estimated independently using the relative validity (RV) methodology, which compares F-ratios for average changes in specific and generic measures across groups differing in the amount of change in the criterion variables and treatment. RV coefficients estimate how each measure responded, relative to the best measure (RV=1.0). An RA-specific measure was based on the Health Assessment Questionnaire (HAQ). Generic measures included eight scales, two summary measures, and an arthritis-specific health index (ASHI) scored from the SF-36 Health Survey. RESULTS: The SF-36 ASHI was most valid (RV = 1.0) for 4 of the 8 criteria. The SF-36 bodily pain (BP) scale was most valid (RV = 1.0) for 2 of the 8 criteria, including treatment. The SF-36 vitality scale was most valid (RV = 1.0) for 1 of the 8 criteria. SF-36 scales measuring physical health were consistently more valid (RV > .30) than scales measuring mental health (RV < .20). The HAQ was highly valid for 2 of the 8 criteria (RV = .89 - .92) and responded moderately well to 4 other criteria (RV = .28 - .49). CONCLUSION: The responsiveness of the arthritis-specific scoring of the SF-36 health profile to changes in severity of rheumatoid arthritis was equal to or better than the generically scored SF-36 health profile and disease specific Health Assessment Questionnaire (HAQ).