Background: During the past decades the number of published randomised clinical trials evaluating the efficacy of conservative treatments for low back pain has substantially increased. At the same time several initiatives have been undertaken in order to improve the methodological quality of published randomised clinical trials, also in the field of low back pain.

Objectives: We investigated how the methodological quality of randomised clinical trials in the field of low back pain developed over time since the first published trial in 1961. Methodological scores of 269 RCTs included in 15 Cochrane reviews of conservative treatment for low back pain were compared.

Results: Overall many trials showed methodological shortcomings. The median overall quality scores ranged from 36% to 82% depending on the type of intervention. There were no improvements in median overall methodological quality over time from 1961 to 2004. However, among non-drug trials the quality improved somewhat from 1995 to present time, and especially when the double blinding criteria (which are usually not feasible in non-drug trials) were omitted from the analysis. Drug trials had comparatively good quality already in the sixties, with no improvement over time, except a small decline in the latter part of nineties.

Conclusions: The overall methodological quality of randomized clinical trials has not improved over the last decades, except somewhat among non-drug trials from 1995. Editors and reviewers of scientific journals should more actively use the Consort-statement in order to improve the quality of future trials.