OBJECTIVE: To determine the effectiveness and predictors of response to lumbar epidural corticosteroid injections (ESI) in patients with sciatica. We performed a 12-month, multicentre, double-blind, randomized, placebo-controlled, parallel-group trial in four secondary pain-care clinics in the Wessex Region. METHODS: Two hundred and twenty-eight patients with a clinical diagnosis of unilateral sciatica of 1-18 months' duration were randomized to either three lumbar ESIs of triamcinolone acetonide or interligamentous saline injections at intervals of 3 weeks. The main outcome measure was the Oswestry low back pain disability questionnaire (ODQ). RESULTS: At 3 weeks, the ESI group demonstrated a transient benefit over the placebo group (patients achieving a 75% improvement in ODQ, 12.5 vs 3.7%; number needed to treat, 11.4). No benefit was demonstrated from 6 to 52 weeks. ESIs did not improve physical function, hasten return to work or reduce the need for surgery. There was no benefit of repeated ESIs over single injection. No clinical predictors of response were found. At the end of the study the majority of patients still had significant pain and disability regardless of intervention. CONCLUSIONS: In this pragmatic study, ESIs offered transient benefit in symptoms at 3 weeks in patients with sciatica, but no sustained benefits in terms of pain, function or need for surgery. Sciatica is a chronic condition requiring a multidisciplinary approach. To fully investigate the value of ESIs, they need to be evaluated as part of a multidisciplinary approach.