The aim of the study was to compare the responsiveness of disease-specific (Arthritis Impact Measurement Scale 2, AIMS2), generic (Medical Outcome Study Short Form Health Survey, SF-36) and preference-based instruments (rating scale, RS and time tradeoff, TTO) to changes in articular status and perceived health in patients with rheumatoid arthritis (RA). Seventy-eight consecutive patients with RA, attending the care facilities of the Department of Rheumatology of Ancona, were recruited in this longitudinal study. In order to assess the responsiveness three strategies were used: effect size (ES), standardised response mean (SRM) and receiver operating characteristic (ROC). There were 55 women and 23 men with a mean age of 56 years (range 19-78) and arthritis duration of 7.1 years (range 6 months to 24 years). Using three-category EULAR criteria as external indicators of improvement/response, 21 patients (27%) reported a significant improvement, 23 (29.5%) moderate improvement, and 34 (43.5%) no change over the 12-month period. The mean change scores in generic and specific health status instruments and utility measures were significantly related to response category. The AIMS2 subscales (physical function, pain, psychological function and social interaction) were slightly more responsive than those of SF-36. The physical and pain dimensions were most sensitive for measuring change over a 12-month period, followed by psychological and social dimensions. For the utility measurement, RS scores were found to be significantly more responsive in detecting changes in preferences than TTO scores. These results may have implications for the application of the health status and utility measures in clinical trials in patients with RA.