OBJECTIVE--To assess the sensitivity to change over time of four health status instruments in relation to patients with rheumatoid arthritis. DESIGN--Observational three month study of four self assessed instruments (arthritis impact measurement scales (AIMS), health assessment questionnaire (HAQ), Nottingham health profile (NHP), functional limitations profile (FLP)). SETTING--One rheumatology unit. PATIENTS--101 patients with definite or classic rheumatoid arthritis. MAIN MEASURES--Change scores for dimensions of instruments, as determined by effect size (mean change in score/baseline standard deviation of variable) and conventional rheumatological measures, at baseline and after three months. RESULTS--Change scores for comparable dimensions (mobility, activities of daily living, household, pain, mood or emotion, and social scales) of the instruments were compared among 30 patients who considered their health status to have improved over three months. For all dimensions of health status the magnitude of change varied considerably according to the instrument. Maximum range in effect size was for social scales (AIMS 0.06, NHP 0.24, FLP 0.60). No single instrument seemed consistently to show the most change over all dimensions. CONCLUSION--Selection of health status instruments for audit or evaluation may have a considerable impact on the pattern of results obtained, and the "responsiveness" of such scales should be as carefully examined as their reliability and acceptability when selecting outcome measures.