The arterial blood ketone body ratio (AKBR: acetoacetate/3-hydroxybutyrate) within 48 h of admission is reported to be an excellent prognostic indicator for acute hepatic failure. In this study, we assessed the AKBR in 63 patients receiving supportive medical therapy for alcoholic hepatitis, in order to investigate its efficacy for predicting complications and the prognosis. Twelve patients (19%) died and 51 patients (81%) survived. Hepatic encephalopathy, severe coagulopathy, and renal failure were the critical complications (P < 0.01), and the AKBR at 72 h of hospitalization was closely correlated with these complications (P < 0.01), although they could not be predicted in any other way during the early admission period. The AKBR of normal individuals ranged from 1.0 to 2.1 (1.54 +/- 0.26, mean +/- SD), so an AKBR > 1.0 (mean-2SD) was defined as normal. The AKBR value at 48 and 72 h of hospitalization showed a significant difference between survivors and non-survivors (P < 0.01). All survivors showed an increase of the AKBR to above 0.7 at 72 h, with subsequent maintenance of the ratio over 1.0, while eight of the 12 non-survivors had sustained suppression of the AKBR below 0.7 at 72 h. Seven of these eight patients subsequently developed multiple organ failure. These findings suggest that the AKBR could be a possible marker of potentially fatal complications and a poor prognosis in patients with alcoholic hepatitis.