Objective: The purposes of this study were to systematically review and apply regression analysis to randomised controlled trials [RCTs] that evaluated the effectiveness of Pilates exercise in improving persistent, non-specific low back pain and functional disability. Methods: Electronic databases were searched from January 1950 to March 2011. Articles were eligible for inclusion if they were RCTs comparing Pilates exercise with a placebo treatment [PT], minimal intervention [MI] or another physiotherapeutic treatment [APT]. Results: Nine trials were included. Pilates was moderately superior to APT (pooled Effect Size [ES] weighted = -0.55, 95% confidence interval [CI] = -0.08 to -1.03) in reducing disability but not for pain relief. Pilates provided moderate to superior pain relief compared to MI (pooled ES weighted = -0.44, 95% CI = -0.09 to -0.80) and a similar decrease in disability. The statistical model used did not detect any predictor variable. Conclusions: Due to the presence of co-interventions and the low methodological quality of some studies, these conclusions should be interpreted with caution.