Seventeen of 41 patients with lupus nephritis who underwent dialysis for renal failure recovered renal function and discontinued dialysis. Two of these 17 had confounding factors unrelated to lupus that contributed to renal dysfunction (one meningococcemia, one vigorous diuresis). Indications for dialysis were identical both in patients who discontinued dialysis (short-term) and in those who did not (long-term). The rate of progression to dialysis, measured as the slope of the reciprocal of the serum creatinine level versus time, was significantly more rapid in the short-term group (p < 0.001). Patients who underwent short-term dialysis were more likely to have had lupus for less than two years (p = 0.015). Anti-DNA antibody binding values, total hemolytic complement levels, extent of extrarenal disease, and hypertension did not differentiate the short-term from long-term dialysis groups. Renal biopsy performed within three months of first dialysis did not demonstrate a consistent picture in the short-term dialysis group. Dialysis is not equivalent to irrevocable end-stage renal disease in patients with lupus nephritis. Thirteen of 22 patients (59 percent) with a 10 percent reduction time for renal function of less than three weeks were able to discontinue dialysis. Ten of these 13 were alive without need for dialysis six months later, with a mean follow-up serum creatinine level of 2.9 ± 1.9 mg/dl.