Several versions of the 24-item Roland Morris Disability Questionnaire (RMDQ) have been proposed; however, their responsiveness has not been extensively explored. The objective of this study was to compare the responsiveness of four versions of the RMDQ. Perceived disability was measured using the 24-item, two 18-item and an 11-item RMDQ on 1,069 low back pain patients from six randomised controlled trials. Responsiveness was calculated using effect size, Guyatt's responsiveness index (GRI) and receiver operating characteristics (ROC) curves. Effect size analyses showed that both 18-item versions of the RMDQ were superior to the 24- and 11-item versions of the RMDQ. GRI showed that the 24- and 18-item versions of the RMDQ were similar but more responsive than the 11-item. ROC curves revealed that the 11-item was less responsive than the other three versions, which had similar responsiveness. The results of this study demonstrate that the 24-item and both 18-item versions of the RMDQ have similar responsiveness with all having superior responsiveness to the 11-item.