AIM: To compare the efficacy of pentoxifylline and prednisolone in the treatment of severe alcoholic hepatitis, and to evaluate the role of different liver function scores in predicting prognosis. METHODS: Sixty-eight patients with severe alcoholic hepatitis (Maddrey score ≥ 32) received pentoxifylline ( n = 34, group Ⅰ) or prednisolone ( n = 34, group Ⅱ) for 28 d in a randomized double-blind controlled study, and subsequently in an open study (with a tapering dose of prednisolone) for a total of 3 mo, and were followed up over a period of 12 mo. RESULTS: Twelve patients in group Ⅱ died at the end of 3 mo in contrast to five patients in group Ⅰ. The probability of dying at the end of 3 mo was higher in group Ⅱ as compared to group Ⅰ (35.29% vs 14.71%, P = 0.04; log rank test). Six patients in group Ⅱ developed hepatorenal syndrome as compared to none in group Ⅰ. Pentoxifylline was associated with a significantly lower model for end-stage liver disease (MELD) score at the end of 28 d of therapy (15.53 Maddrey score was associated with increased mortality. CONCLUSION: Reduced mortality, improved risk-benefit profile and renoprotective effects of pentoxifylline compared with prednisolone suggest that pentoxifylline is superior to prednisolone for treatment of severe alcoholic hepatitis.