Nephritis in Henoch-Schönlein purpura (HSP) is the primary cause of morbidity and mortality. Although many therapeutic regimens have been reported to be effective, no therapy has been shown in a controlled trial to be beneficial. Fifty-six patients with histopathologically severe HSP nephritis were randomized to receive supportive therapy with or without cyclophosphamide, 90 mg/m(2)/day for 42 days. Patients were classified according to status at final follow-up: Fully Recovered 48.2%, Persistent Abnormalities 39.3%, or ESRD/Death 12.5%. There were no differences in onset data or outcome between the two trial groups or in outcome between trial and 23 non-trial patients followed concurrently. Therefore, data from trial and non-trial patients were combined for further analysis. There was no correlation between outcome and age, blood pressure, serum total protein, or serum albumin. Although rates of proteinuria did not correlate with outcome, all those with progression to ESRD had nephrotic levels of proteinuria at onset. Only five of 28 patients with nephrotic levels of proteinuria and severe onset histopathology recovered fully. No patient with crescents in 50% or more of glomeruli went on to full recovery. Recurrence of non-renal symptoms did not correlate with outcome. Nephrotic syndrome, decreased GFR, and more severe histopathology at onset, as well as persistence of urinary abnormalities for several years, are ominous signs.