To assess the responsiveness of the three most used patient reported multiple sclerosis (MS) specific questionnaires: the Functional Assessment of MS (FAMS), the MS Impact Scale (MSIS-29) and the 54 item MS Quality of Life (MSQOL-54). Prospective multicentre longitudinal study on 104 MS patients treated with intravenous steroids for clinical exacerbation. Patient reported data, Expanded Disability Status Scale (EDSS) score and clinical information were collected at admission and 8 weeks later. "Internal" (distribution based) responsiveness was assessed by standardised response means (SRM). "External" (anchor based) responsiveness was assessed by receiver operating characteristic (ROC) curves in relation to corresponding changes in a pre-specified reference measure (anchor). The pre-specified anchor was patients' self-reported recovery assessed on a 5 point Likert scale. SRM was 0.39 for FAMS, 0.58 for MSIS-29 physical scale, 0.45 for MSIS-29 psychological scale, 0.71 for MSQOL-54 physical health composite and 0.57 for MSQOL-54 mental health composite. Seventy-three patients (70%) reported they had improved; physicians agreed substantially with patient assessments (kappa statistic 0.70, 95% CI 0.54 to 0.85). Areas under ROC curves differed significantly from 0.50 only for the MSIS-29 and MSQOL-54 scales where areas ranged from 0.65 (95% CI 0.53 to 0.76) for the MSIS-29 psychological scale to 0.70 (95% CI 0.58 to 0.81) for the MSQOL-54 mental health composite. Areas under ROC curves assessed using a physician based anchor were similar to the patient based areas. The responsiveness of the MS specific instruments was less than ideal. The MSIS-29 and MSQOL-54 were significantly more responsive, using both distribution based and anchor based approaches, than FAMS, and should be preferred in longitudinal studies.