Anti-glomerular basement membrane (GBM) disease usually presents with rapidly progressive glomerulonephritis accompanied by pulmonary hemorrhage. The low incidence and fulminant course of disease preclude a large randomized controlled study to define the benefits of any given therapy. We conducted a retrospective survey of 221 consecutive patients seen from 1998 to 2008 in our hospital, and report here the patient and renal survival and the risk factors affecting the outcomes. Considering the similar clinical features of the patients, we could compare the effects of 3 different treatment regimens: 1) combination therapy of plasmapheresis and immunosuppression, 2) steroids and cytotoxic agents, and 3) steroids alone.The patient and renal survival rates were 72.7% and 25.0%, respectively, at 1 year after disease presentation. The serum level of anti-GBM antibodies (increased by 20 U/mL; hazard ratio [HR], 1.16; p = 0.009) and the presentation of positive antineutrophil cytoplasmic antibodies (ANCA) (HR, 2.18; p = 0.028) were independent predictors for patient death. The serum creatinine at presentation (doubling from 1.5 mg/dL; HR, 2.07; p < 0.001) was an independent predictor for renal failure.The combination therapy of plasmapheresis plus corticosteroids and cyclophosphamide had an overall beneficial effect on both patient survival (HR for patient mortality, 0.31; p = 0.001) and renal survival (HR for renal failure, 0.60; p = 0.032), particularly patient survival for those with Goodpasture syndrome (HR for patient mortality, 0.29; p = 0.004) and renal survival for those with anti-GBM nephritis with initial serum creatinine over 6.8 mg/dL (HR for renal failure, 0.52; p = 0.014). The treatment with corticosteroids plus cyclophosphamide was found not to improve the renal outcome of disease (p = 0.73). In conclusion, the combination therapy was preferred for patients with anti-GBM disease, especially those with pulmonary hemorrhage or severe renal damage. Early diagnosis was crucial to improving outcomes.