A meta-analysis was undertaken of low-level laser therapy (LLLT) on musculoskeletal pain. A literature search revealed 23 LLLT trials and of these 17 were controlled trials. Ten were double blind and 7 were insufficiently blinded. Within the studies identified pain was assessed by visual analogue scale or by "some other indices of pain". Nine double-blind trials and 4 controlled trials presented results in a form which allowed pooling of data. In the double-blind trials, the mean difference in pain between LLLT and placebo was 0.3% (S.E.(d) 4.6%, confidence limits -10.3-10.9%). In the insufficiently blinded trials the mean difference in pain was 9.5% (S.E.(d) 4.5%, confidence limits -2.9-21.8%). We conclude that LLLT has no effect on pain in musculoskeletal syndromes.