The value of epidural injections of corticosteroid as an outpatient treatment of sciatica has been hitherto uncertain. An epidural injection of 80 mg methylprednisolone in 10 ml physiological saline was compared with an interspinous injection of 2 ml physiological saline in a double blind fashion amongst 39 outpatients. Significant differences of pain relief were seen between the two groups within 2 weeks. This benefit disappeared for six (35%) patients within 6 months of treatment although 11 (65%) successfully treated subjects had sustained improvement up to this time. Outpatient epidural injections of corticosteroid are thus a useful short-term means of relieving pain in sciatica but probably have little effect on the long-term natural history of symptoms. Factors associated with a failure to respond to epidural steroid injections are discussed.