To determine the potential steroid sparing effect of plasma exchange in pemphigus we enrolled 40 patients in a multicenter randomized study. Eighteen patients were treated by prednisolone alone, 22 by prednisolone plus ten large-volume plasma exchanges over four weeks. All patients received oral prednisolone in the same initial dosage (0.5 mg/kg/d), which was increased weekly if needed. The number of cases controlled at each therapeutic step did not differ between the two groups. In eight cases, four in each group, the disease was not controlled by the highest therapeutic step of the protocol, with four deaths from sepsis in the plasma exchange group. The controlled cases needed similar cumulative prednisolone doses (5237 +/- 5512 mg in the plasma exchange group vs 4246 +/- 1601 mg in the control group). The evolution of serum pemphigus antibody was not different in the two groups. These findings suggest that plasma exchange in association with low steroid doses alone are not effective in the treatment of pemphigus and may even promote sepsis.