The efficacy of epidural corticosteroids in the treatment of sciatica was investigated by meta-analysis of all randomized controlled trials. Eleven suitable trials of good quality were identified involving a total of 907 patients. The use of epidural (caudal or lumbar) steroid in the short-term (up to 60 days) increased the odds ratio (OR) of pain relief ( > 75% improvement) to 2.61 (95% CI 1.90-3.77) when compared with placebo. Despite some variations in trial characteristics there was little evidence of significant heterogeneity (P = 0.07). When the trials were analysed for near or total relief of pain in the short-term the OR is 2.79 (95% CI 1.92-4.06), for heterogeneity (P = 0.07). For longterm relief of pain (up to 12 months) the OR is 1.87 (95% CI 1.31-2.68). Efficacy is independent of the route of injection; for caudal epidural steroid the OR is 3.80 (95% CI 1.36-10.6) and for the lumbar epidural steroid 2.43 (95% CI 1.77-3.74). Adverse events included dural tap (2.5%), transient headache (2.3%) and a transient increase in pain (1.9%). There were no reported longterm adverse events. In conclusion we present quantitative evidence from meta-analysis of pooled data from randomized trials that epidural administration of corticosteroids is effective in the management of lumbosacral radicular pain.