Aims: The short-term mortality of severe alcoholic hepatitis (SAH) is very high, but there are no effective treatments to improve short-term mortality other than corticosteroid. This study investigated the effects of rifaximin treatment in patients with SAH. Methods: In an open-label trial, patients with SAH (Maddrey’s discriminant function≥32) were randomized to rifaximin or control group, each added to corticosteroid or pentoxifylline for 4 weeks. Randomization was stratified by SAH treatment. Liver transplantation free survival was evaluated. (NCT02485106) Results: Total 49 patients were enrolled in this study (29 in control and 20 in rifaximin group). The mean Model for End-stage Liver Disease (MELD) score were 24.4 and 27.8 in control and rifaximin group (P=0.083). Rifaximin treatment was tolerable and only 1 patients stopped due to adverse event. There were no differences in 3-month and 6-month mortality between two groups (P=0.576 and P=0.239, respectively). Corticosteroid group had higher 3-month and 6-month survival than pentoxifylline group (P=0.03 and P=0.016, respectively). When stratified by SAH treatment, there were no significant 3-month and 6-month survival between control and rifaximin treatment (P=0.516 and P=0.937 in corticosteroid group and P=0.948 and P=0.620 in pentoxifylline group, respectively). Cox Proportional hazard model showed that MELD score, white blood cell count, C-reactive protein were significant factors for 6-month survival, and MELD score was only independent factor for 6-month survival (Hazard ratio 1.188, P=0.001). Conclusions: In patients with SAH, adding rifaximin to corticosteroid or pentoxifylline was tolerable but had no survival benefit. MELD score was only significant factor for short-term mortality.