Plasmapheresis combined with immunosuppression dramatically improved the survival of patients with Goodpasture's disease in the late 1970's. The presence of circulating pathogenic autoantibodies in this disease provided a logical rationale for the use of plasma exchange therapy. Careful analysis of the response to treatment has suggested that patients presenting with a serum creatinine < 600 mumol/I have the most benefit from plasma exchange. Subsequently plasmapheresis has been tried in a variety of other nephritides, predominantly those causing rapidly progressive glomerulonephritis, with variable success. Initial studies failed adequately to distinguish patients with a number of quite distinct causes of crescentic nephritis. However it has become clear that plasma exchange significantly improves the outcome of patients with pauci-immune crescentic glomerulonephritis who present with severe renal failure requiring dialysis, but not those with less severe renal disease.