The objective of this study was to compare the responsiveness and assess the concurrent validity of two functional health status instruments, the Dartmouth COOP charts and the SF-36 in chronic low-back pain (CLBP) patients. The data came from 129 of 174 patients who participated in a randomized clinical trial of the therapeutic management of CLBP. Reliable and valid disease-specific outcomes, patient-rated low-back pain and disability, were used as external criteria (EC) to identify improved and non-improved patients. Unpaired t-statistics and receiver operating characteristic (ROC) curve calculations were used to quantify responsiveness. The two instruments had sufficient and very similar responsiveness using both EC. Comparisons between improved and non-improved patients for the COOP charts and SF-36, respectively, using pain as EC, yielded differences which translated into large effect sizes (0.8 and 0.7) (P=0.0008 and 0.003). Using disability as EC, differences of moderate effect size were found (0.5 and 0.6) (P=0.02 and 0.002). The ROC curve calculations using pain as EC resulted in areas under the curve of 0.76 (95% CI: 0.64, 0.88) for the COOP charts, and 0.74 (95% CI: 0.60, 0.88) for the SF-36. The corresponding areas using disability as EC were 0.67 (95% CI: 0.55, 0.79) and 0.72 (95% CI: 0.60, 0.84). The best cut-off point in both instruments for differentiating between improved and non-improved patients was approximately six percentage points. The constructs of functional health status, as reflected in the global scores of the two instruments, are highly correlated (r=0.82). Six of the instruments' nine dimensions are moderately to highly correlated (r=0.52 to 0.86), and the overall canonical correlation was high (R=0.9). In conclusion, both instruments seem equally suitable for use as outcome measures in clinical trials on CLBP. The COOP charts are faster to fill out and score.