Twenty-seven patients with lupus nephritis and nephrotic syndrome had persistent disease activity despite an adequate trial of corticosteroids and immunosuppressive drugs; 30% were Asians, compared with 7% of our overall SLE population. Two years later, seven had a very good outcome and seven a poor outcome. Thirty clinical, pathological, laboratory, and treatment variables were analyzed in a good versus poor responder subset comparison in an effort to determine which factors were associated with favorable outcome. Administration of pulse steroids (P = .069) and a low biopsy chronicity index (P = .048) were associated with the good responder subset. Serum creatinine, biopsy class, blood pressure, complement, and anti-DNA values at entry as well as the choice of immunosuppressive drug were not helpful in predicting outcome. All seven good responders were plasmapheresed (P = .026). Patients with refractory lupus nephritis who have a low biopsy chronicity index may benefit from the use of pulse steroids or plasmapheresis, and controlled studies are suggested.