Background: Lumbosacral radicular pain (commonly called sciatica) is a syndrome involving patients who report radiating leg pain. Epidural corticosteroid injections deliver a corticosteroid dose into the epidural space, with the aim of reducing the local inflammatory process and, consequently, relieving the symptoms of lumbosacral radicular pain. This Cochrane Review is an update of a review published in Annals of Internal Medicine in 2012. Some placebo-controlled trials have been published recently, which highlights the importance of updating the previous review. Objectives: To investigate the efficacy and safety of epidural corticosteroid injections compared with placebo injection on pain and disability in patients with lumbosacral radicular pain. Search methods: We searched the following databases without language limitations up to 25 September 2019: Cochrane Back and Neck group trial register, CENTRAL, MEDLINE, Embase, CINAHL, PsycINFO, International Pharmaceutical Abstracts, and two trial registers. We also performed citation tracking of included studies and relevant systematic reviews in the field. Selection criteria: We included studies that compared epidural corticosteroid injections of any corticosteroid drug to placebo injections in patients with lumbosacral radicular pain. We accepted all three anatomical approaches (caudal, interlaminar, and transforaminal) to delivering corticosteroids into the epidural space. We considered trials that included a placebo treatment as delivery of an inert substance (i.e. one with no pharmacologic activity), an innocuous substance (e.g. normal saline solution), or a pharmacologically active substance but not one considered to provide sustained benefit (e.g. local anaesthetic), either into the epidural space (i.e. to mimic epidural corticosteroid injection) or adjacent spinal tissue (i.e. subcutaneous, intramuscular, or interspinous tissue). We also included trials in which a local anaesthetic with a short duration of action was used as a placebo and injected together with corticosteroid in the intervention group. Data collection and analysis: Two authors independently performed the screening, data extraction, and 'Risk of bias' assessments. In case of insufficient information, we contacted the authors of the original studies or estimated the data. We grouped the outcome data into four time points of assessment: immediate (≤ 2 weeks), short term (> 2 weeks but ≤ 3 months), intermediate term (> 3 months but < 12 months), and long term (≥ 12 months). We assessed the overall quality of evidence for each outcome and time point using the GRADE approach. Main results: We included 25 clinical trials (from 29 publications) investigating the effects of epidural corticosteroid injections compared to placebo in patients with lumbosacral radicular pain. The included studies provided data for a total of 2470 participants with a mean age ranging from 37.3 to 52.8 years. Seventeen studies included participants with lumbosacral radicular pain with a diagnosis based on clinical assessment and 15 studies included participants with mixed duration of symptoms. The included studies were conducted mainly in North America and Europe. Fifteen studies did not report funding sources, five studies reported not receiving funding, and five reported receiving funding from a non-profit or government source. Eight trials reported data on pain intensity, 12 reported data on disability, and eight studies reported data on adverse events. The duration of the follow-up assessments ranged from 12 hours to 1 year. We considered eight trials to be of high quality because we judged them as having low risk of bias in four out of the five bias domains. We identified one ongoing trial in a trial registry. Epidural corticosteroid injections were probably slightly more effective compared to placebo in reducing leg pain at short-term follow-up (mean difference (MD) -4.93, 95% confidence interval (CI) -8.77 to -1.09 on a 0 to 100 scale; 8 trials, n = 949; moderate-quality evidence (downgraded for risk of bias)). For disability, epidural corticosteroid injections were probably slightly more effective compared to placebo in reducing disability at short-term follow-up (MD -4.18, 95% CI -6.04 to -2.17, on a 0 to 100 scale; 12 trials, n = 1367; moderate-quality evidence (downgraded for risk of bias)). The treatment effects are small, however, and may not be considered clinically important by patients and clinicians (i.e. MD lower than 10%). Most trials provided insufficient information on how or when adverse events were assessed (immediate or short-term follow-up) and only reported adverse drug reactions - that is, adverse events that the trialists attributed to the study treatment. We are very uncertain that epidural corticosteroid injections make no difference compared to placebo injection in the frequency of minor adverse events (risk ratio (RR) 1.14, 95% CI 0.91 to 1.42; 8 trials, n = 877; very low quality evidence (downgraded for risk of bias, inconsistency and imprecision)). Minor adverse events included increased pain during or after the injection, non-specific headache, post-dural puncture headache, irregular periods, accidental dural puncture, thoracic pain, non-local rash, sinusitis, vasovagal response, hypotension, nausea, and tinnitus. One study reported a major drug reaction for one patient on anticoagulant therapy who had a retroperitoneal haematoma as a complication of the corticosteroid injection. Authors' conclusions: This study found that epidural corticosteroid injections probably slightly reduced leg pain and disability at short-term follow-up in people with lumbosacral radicular pain. In addition, no minor or major adverse events were reported at short-term follow-up after epidural corticosteroid injections or placebo injection. Although the current review identified additional clinical trials, the available evidence still provides only limited support for the use of epidural corticosteroid injections in people with lumbosacral radicular pain as the treatment effects are small, mainly evident at short-term follow-up and may not be considered clinically important by patients and clinicians (i.e. mean difference lower than 10%). According to GRADE, the quality of the evidence ranged from very low to moderate, suggesting that further studies are likely to play an important role in clarifying the efficacy and tolerability of this treatment. We recommend that further trials should attend to methodological features such as appropriate allocation concealment and blinding of care providers to minimise the potential for biased estimates of treatment and harmful effects.