BACKGROUND The relative efficacy of plasma exchange (PE) and intravenous immunoglobulin (IVIg) for the treatment of Guillain-Barré syndrome has not been established. We compared PE with IVIg, and with a combined regimen of PE followed by IVIg, in an international, multicentre, randomised trial of 383 adult patients with Guillain-Barré syndrome. METHODS The patients were randomly assigned PE (five 50 mL/kg exchanges over 8-13 days), IVIg (Sandoglobulin, 0.4 g/kg daily for 5 days), or the PE course immediately followed by the IVIg course. The inclusion criteria were severe disease (aid needed for walking) and onset of neuropathic symptoms within the previous 14 days. Patients were followed up for 48 weeks. RESULTS Four patients were excluded because they did not meet the randomisation criteria. All the remaining 379 patients were assessed for the major outcome criterion-change on a seven-point disability grade scale-by an observer unaware of treatment assignment, 4 weeks after randomisation. At that time, the mean improvement was 0.9 (SD 1.3) in the 121 PE-group patients, 0.8 (1.3) in the 130 IVIg-group patients, and 1.1 (1.4) in the 128 patients who received both treatments (intention-to-treat analysis). None of the differences between the groups for this major outcome criterion was significant. The difference between PE alone and IVIg alone was so small that a 0.5 grade difference was excluded at the 95% level of confidence. There was no significant difference between any of the treatment groups in the secondary outcome measures: time to recovery of unaided walking, time to discontinuation of ventilation, and trend describing the recovery from disability up to 48 weeks. There was a non-significant trend towards a more favourable outcome on some outcome measures with combined treatment. CONCLUSIONS In treatment of severe Guillain-Barré syndrome during the first 2 weeks after onset of neuropathic symptoms, PE and IVIg had equivalent efficacy. The combination of PE with IVIg did not confer a significant advantage.