- Goals: To evaluate the role of the discriminant factor in predicting mortality and deciding on treatment in acute alcoholic hepatitis. Background: Current guidelines on the treatment of alcoholic hepatitis restrict the use of corticosteroids to patients with discriminant factor > 32 (severe disease) because of the toxicity of steroids. Less toxic forms of therapy, such as proxyphylline, may have a role in patients with lower discriminant factor, if mortality without therapy is common. Study: We performed a 5-year retrospective analysis comparing the outcomes of patients with mild and severe alcoholic hepatitis. Receiver operator characteristic curves were used to study the accuracy of the discriminant factor to predict short-term mortality. Results: Among the 41 patients with severe alcoholic hepatitis (discriminant factor > 32) and 48 with mild alcoholic hepatitis, 16 (39%) and 8 (16.7%), respectively, died within 28 days of admission. Only 11 (32%) actually received corticosteroid therapy. The sensitivity and specificity of the discriminant factor in predicting mortality was 66.7% and 61.5%, respectively. A receiver operator characteristic curve of the discriminant factor gave the optimal value for the discriminant factor as 33, with the area under the curve being 0.666 (P = 0.0078; 95% CI = 0.531–0.801). Conclusions: Using the value of around 32 maximizes sensitivity and specificity of the discriminant factor in predicting mortality in alcoholic hepatitis. However, there is a high mortality in patients with alcoholic hepatitis and a discriminant factor less than 32. Alternative effective agents should be considered in patients with milder alcoholic hepatitis.