Appropriate attention can be directed to quality of life indices in the evaluation of therapeutic interventions only if reliable, valid measurement tools with good sensitivity to change are available. The goal of this study was to validate the French version of the Arthritis Impact Measurement Scales 2 (AIMS2), called EMIR (Echelle de Mesure de l'Impact de la polyarthrite Rhumatoïde) and to compare it with the validated French version of the Nottingham Health Profile, called ISPN (Indicateur de Santé Perceptuelle de Nottingham). The French version of the AIMS2 (37 items, 12 dimensions) was obtained via several independent translations and back-translations, followed by selection of the most relevant items by a panel of experts and a preliminary evaluation in rheumatoid arthritis patients. The measurement properties of EMIR (reliability, validity and sensitivity to change) were investigated in a cohort of rheumatoid arthritis patients who were put under methotrexate therapy and followed up for six months. Reliability was evaluated by test-retest at a ten-day interval (intraclass coefficients of correlation). Principal component factorial analysis was used to study construct validity and Cronbach's alpha coefficients to study internal consistency. Convergent validity was evaluated based on correlations between the quality of life scores obtained in selected dimensions of the EMIR and a number of other parameters (number of painful/swollen joints, pain severity score on a visual analog scale, erythrocyte sedimentation rate). Sensitivity to change after three and six months was determined by calculating mean standardized response means. The EMIR and ISPN were compared based on scores and sensitivity to change for the dimensions that investigated the same concepts. One hundred twenty-seven rheumatoid arthritis patients scheduled for methotrexate therapy were entered into the study. Principal component analysis established that all dimensions of the EMIR were independent from one another, except the "walking and bending" dimension. Internal consistency was satisfactory for each of the 12 dimensions, with Cronbach's alpha coefficients ranging from 0.70 to 0.90. Most correlations between quality of life scores and clinical or laboratory parameters were significant, indicating satisfactory convergent validity. The reliability study also yielded satisfactory results, with intraclass coefficients ranging from 0.65 to 0.90. Sensitivity to change after three and six months was significant for 11 of the 12 dimensions (mean standardized responses, 0.30 to 0.77). Sensitivity to change was slightly better for the EMIR than for the ISPN. Analysis of scores demonstrated that these two instruments did not measure quality of life in exactly the same way but complemented each other.