In a multicentre, randomised trial of prednisolone in acute polyneuropathy of undetermined aetiology (Guillain-Barré syndrome), 21 patients were treated with prednisolone (60 mg daily for one week, 40 mg daily for four days, and then 30 mg daily for three days) and 19 did not have steroid treatment. Patients were graded on a six-point scale by one of two neurologists who had no knowledge of the treatment schedule. Reassessment at one, three, and twelve months consistently showed greater improvement in the control than the prednisolone group but the only statistically significant result was in the improvement at three months among patients entered to the trial within a week of onset of illness. The 6 control patients had improved by 2.5 +/- 0.43 grades by three months from entry to the trial whereas the 10 prednisolone patients had only improved by 0.9 +/- 0.46 grades (P less than 0.05). There was 1 death related to the polyneuropathy in each group, and 1 suicide in a control patient during convalescence. 6 prednisolone patients were left with considerable disability compared with 1 control patient. There were 3 relapses in the prednisolone group, but none in the control group. The results indicate that steroid treatment is not beneficial and can be detrimental in acute neuropathy of undetermined aetiology.