Most patients with acute inflammatory polyneuropathy (AIP) recover spontaneously, but the time course of the illness is unpredictable so that the results of treatment are difficult to assess. Three decades of retrospective reports of steroid treatment fail to demonstrate any striking beneficial effect. In a randomized trial of prednisolone, starting dose 60 mg daily, 21 treated patients improved more slowly than 19 untreated patients. By contrast, in rats immunized with bovine nerve root myelin, prednisolone at 10 mg/kg reduced the severity and duration of experimental allergic neuritis (EAN), the putative animal model for AIP. This discrepancy might reflect the greater difficulty of clinical as opposed to animal therapeutic trials or indicate that EAN is not the appropriate model for the human disease. Immunosuppressive drugs, plasmapheresis and other agents have also been employed, but their efficacy cannot be decided from the available case report. The role of similar agents in chronic progressive and relapsing inflammatory neuropathy cannot yet be resolved, but in some patients steroids do appear to be valuable.