Several investigators have given priority to treatment of AH, as this entity is associated with significant early mortality. Patients at significant risk of early death are identified by the use of prognostic models. The Lille model highlights the benefits obtained from strategy integrating the impact of treatment upon the evaluated endpoint. Survival impact of corticosteroids is significant in complete and partial responders, whereas it appeared negligible in null-responders. Infection screening is warranted, but should not contraindicate steroids. In terms of mechanisms, non-response to steroids is the key factor in development of infection and prediction of survival. For patients with severe alcoholic hepatitis failing to medical therapy can be identified earlier and have a 6-month survival around 30%; early liver transplantation in those patients is attractive but highly controversial as it challenges the 6-month abstinence rule prior to LT.