We carried out a prospective randomized trial comparing pulse cyclophosphamide and pulse methylprednisolone in 29 patients with severe lupus nephritis in activity. Patients were assigned to one of two regimens: monthly pulse cyclophosphamide (0.5-1.0 g/m2 body surface area) for 4 months, followed by bimonthly doses for 4 months and quarterly doses for 6 months (14 patients) or pulse methylprednisolone (10-20 mg/kg weight) initially for 3 consecutive days and thereafter in the same intervals as the alternative regimen (15 patients). The mean follow-up was 15 months. Two patients in the cyclophosphamide group and three in the methylprednisolone group died. Renal failure (doubling of serum creatinine) developed in four patients in the cyclophosphamide group compared with five patients in the methylprednisolone group. Cumulative probability of not doubling serum creatinine was similar for cyclophosphamide and methylprednisolone groups (0.66 vs 0.69, respectively, P > 0.20, after 18 months). Cumulative probability of survival without renal failure was also not significantly different (0.61 and 0.63, respectively, P > 0.20, after 18 months). These results suggest that pulse cyclophosphamide is as effective as pulse methylprednisolone in preserving renal function in patients with severe lupus nephritis.