Sixty-three patients with crescentic glomerulonephritis (cellular crescents in greater than 50% of glomeruli) were considered for a prospective randomized trial comparing intravenous methylprednisolone, prednisone, and azathioprine with and without plasma exchange. Of 32 patients who fulfilled the inclusion criteria for this study, 16 were randomly assigned to receive drug therapy (control) and 16 to receive plasma exchange as well. The randomization was stratified for initial need of dialysis, and the presence of oliguria and sclerosis. Renal pathology was similar in the two groups of patients. There was no significant difference in the number of patients initially on dialysis who were able to discontinue it during the study (2/7 control v 3/4 plasma exchange), whereas no control but two plasma exchange-treated patients started dialysis during the study. Serum creatinine at randomization was similar in the two groups: 769 +/- 486 mumol/L (8.7 +/- 5.5 mg/dL) in the control group versus 643 +/- 275 mumol/L (7.3 +/- 3.1 mg/dL) in the plasma exchange group. There was no significant difference between the two groups in mean serum creatinine, change in serum creatinine, change in reciprocal, or change in logarithm of serum creatinine at 1, 3, 6, or 12 months following randomization. Power calculation, assuming a 20% difference would be clinically relevant, was 0.94 at 12 months. There was significant morbidity in both groups; there were two deaths within 1 year of randomization, both of pulmonary infection and both in the plasma exchange group. We conclude that plasma exchange offers no additional therapeutic benefit to patients with idiopathic rapidly progressive glomerulonephritis (RPGN) who are not dialysis-dependent at presentation.