The purpose of this prospective, single site cohort quasi-experimental study was to determine the responsiveness of the numerical rating scale (NRS), Roland-Morris disability questionnaire (RMDQ), Oswestry disability index (ODI), pain self-efficacy questionnaire (PSEQ) and the patient-specific functional scale (PSFS) in order to determine which would best measure clinically meaningful change in a chronic low back pain (LBP) population. Several patient-based outcome instruments are currently used to measure treatment effect in the chronic LBP population. However, there is a lack of consensus on what constitutes a "successful" outcome, how an important improvement/deterioration has been defined and which outcome measure(s) best captures the effectiveness of therapeutic interventions for the chronic LBP population. Sixty-three consecutive patients with chronic LBP referred to a back exercise and education class participated in this study; 48 of the 63 patients had complete data. Five questionnaires were administered initially and after the 5-week back class intervention. Also at 5 weeks, patients completed a global impression of change as a reflection of meaningful change in patient status. Score changes in the five different questionnaires were subjected to both distribution- and anchor-based methods: standard error of measurement (SEM) and receiver operating characteristic (ROC) curves to define clinical improvement. From these methods, the minimal clinically important difference (MCID) defined as the smallest difference that patients and clinicians perceive to be worthwhile is presented for each instrument. Based on the SEM, a point score change of 2.4 in the NRS, 5 in the RMDQ, 17 in the ODI, 11 on the PSEQ, and 1.4 on the PSFS corresponded to the MCID. Based on ROC curve analysis, a point score change of 4 points for both the NRS and RMDQ, 8 points for the ODI, 9 points for the PSEQ and 2 points for the PSFS corresponded to the MCID. The ROC analysis demonstrated that both the PSEQ and PSFS are responsive to clinically important change over time. The NRS was found to be least responsive. The exact value of the MCID is not a fixed value and is dependent on the assessment method used to calculate the score change. Based on ROC curve analysis the PSFS and PSEQ were more responsive than the other scales in measuring change in patients with chronic LBP following participation in a back class programme. However, due to the small sample size, the lack of observed worsening of symptoms over time, the single centre and intervention studied these results which need to be interpreted with caution.