Corticosteroids are usually given for management of Graves' ophthalmopathy, but they have many and serious side-effects. By comparison, retrobulbar irradiation is well tolerated, although its efficacy has been evaluated only in uncontrolled studies. Therefore, we did a double-blind randomised trial, in which 28 patients with moderately severe Graves' ophthalmopathy were treated with a 3-month course of oral prednisone and sham irradiation, and 28 received retrobulbar irradiation (20 Gy) and placebo capsules. Therapeutic outcome, assessed twenty-four weeks after the start of treatment, was determined by the change in the highest NOSPECS class. A successful outcome was observed in 14 prednisone-treated and in 13 irradiated patients. Responders to treatment (but not nonresponders) in both groups showed improvements in total and subjective eye score and a decrease in eye-muscle volume. Response to either treatment was due largely to changes in soft-tissue involvement and eye-muscle motility. Mean elevation in responders to radiotherapy increased from 18.5 degrees (95% CI 14.8-22.2) at baseline to 21.8 degrees (18.6-25.0) at week twenty-four (p = 0.003), but did not change in prednisone responders. Side-effects were more frequent and severe during prednisone than during radiotherapy. Radiotherapy and oral prednisone appear to be equally effective as initial treatment in patients with moderately severe Graves' ophthalmopathy. In view of its better tolerability, radiotherapy should be considered the treatment of first choice.