In patients who have anti-neutrophil cytoplasm autoantibody (ANCA)-associated glomerulonephritis and are on dialysis at time of diagnosis, renal function is sometimes insufficiently restored by immunosuppressive treatment, which often coincides with potentially lethal adverse effects. This study investigated the clinical and histologic variables that determine the chances of dialysis independence, dialysis dependence, or death after 12 mo in these patients. Sixty-nine patients who had ANCA-associated glomerulonephritis and were dialysis dependent at diagnosis received uniform, standard immunosuppressive therapy plus either intravenous methylprednisolone or plasma exchange. Eleven clinical and histologic variables were assessed. Univariate and binary logistic regression analyses were performed. Predictive parameters were entered into a two-step binary logistic regression analysis to differentiate among the outcomes of dialysis independence, dialysis dependence, or death. The point at which the chance of therapy-related death exceeded the chance of dialysis independence was determined. The chance of recovery exceeded the chance of dying in most cases. Intravenous methylprednisolone as adjunctive therapy plus <18% normal glomeruli and severe tubular atrophy increased the chance of therapy-related death over the chance of dialysis independence. Plasma exchange treatment plus severe tubular atrophy and <2% normal glomeruli increased the chance of therapy-related death over that of dialysis independence. Even with ominous histologic findings, the chance of renal recovery exceeds the chance of therapy-related death when these patients are treated with plasma exchange as adjunctive therapy.