A controlled randomized trial of plasma exchange combined with prednisone was compared with supportive care alone in patients with Guillain-Barré syndrome (GBS). There was no significant improvement in the treated group over the controls. The sample size, albeit small (12 treated and 13 controls), had the power (95% chance) to detect a change of 2 British Medical Research Council grades of strength between the groups. The difference in our results and others that indicated a beneficial effect of plasma exchange may reflect an adverse effect of prednisone on recovery in GBS. Failure to find a benefit of plasma exchange in our patients also raises the possibility that GBS may be a heterogeneous disorder with humoral factors playing a role in some but not all patients.