Alcoholic hepatitis (AH) is an acute inflammatory syndrome causing significant morbidity and mortality. The prognosis is strongly dependent on disease severity, as assessed by clinical scoring systems. Reliable epidemiological data as well as knowledge of the clinical course of AH are essential for planning and resource allocation within the health care system. Likewise, individual evaluation of risk is desirable in the clinical handling of patients with AH as it can guide treatment, improve patient information, and serve as strata in clinical trials. The present PhD thesis is based on three studies using a cohort of nearly 2000 patients diagnosed with AH in Denmark from 1999 to 2008 as a cohort, in a population-based study design. The aims of this thesis were as follows. (1) To describe the incidence and short- and long-term mortality, of AH in Denmark (Study I). (2) To validate and compare the ability of the currently available prognostic scores to predict mortality in AH (Study II). (3) To investigate the short- and long-term causes of death of patients with AH (Study III). During the study decade, the annual incidence rate in the Danish population rose from 37 to 46 per 106 for men and from 24 to 34 per 106 for women. Both short- and long-term mortality rose for men and women, and the increase in short-term mortality was attributable to increasing patient age and prevalence of cirrhosis. Our evaluation of the most commonly used prognostic scores for predicting the mortality of patients with AH showed that all scores performed similarly, with Area under the Receiver Operator Characteristics curves giving values between 0.74 and 0.78 for 28-day mortality assessed on admission. Our study on causes of death showed that in the short-term (< 84 days after diagnosis), patients with AH were likely to die from liver-related events and infections. In the long-term (≥ 84 days after diagnosis), those who developed cirrhosis mainly died from liver-related causes, and those who did not develop cirrhosis mainly died from causes related to alcohol abuse. In conclusion, the present thesis provides novel warranted epidemiological information about AH that shows increasing incidence and mortality rates. Consequently, it reiterates the fact that AH is a life-threatening disease and suggests that AH is an increasing public health concern. The most widely used prognostic models may be helpful adjuvants in the routine management of patients with AH, provided that clinicians are aware of the models' limitations. The causes of death in AH are primarily due to liver-related complications, suggesting that patients with AH could benefit from continued follow-up by a hepatologist after the acute episode.