Alcoholic hepatitis (AH) is a distinct presentation of alcoholic liver disease arising in patients who have been drinking to excess for prolonged periods, which is characterised by jaundice and liver failure. Severe disease is associated with high short-term mortality. Prednisolone and pentoxifylline (PTX) are recommended in guidelines for treatment of severe AH, but trials supporting their use have given heterogeneous results and controversy persists about their benefit. The aim of the clinical effectiveness and cost-effectiveness of STeroids Or Pentoxifylline for Alcoholic Hepatitis trial was to resolve the clinical dilemma on the use of prednisolone or PTX. The trial was a randomised, double-blind, 2 × 2 factorial, multicentre design. Sixty-five gastroenterology and hepatology inpatient units across the UK. Patients with a clinical diagnosis of AH who had a Maddrey's discriminant function value of ≥ 32 were randomised into four arms: A, placebo/placebo; B, placebo/prednisolone; C, PTX/placebo; and D, PTX/prednisolone. Of the 5234 patients screened for the trial, 1103 were randomised and after withdrawals, 1053 were available for primary end-point analysis. Those allocated to prednisolone were given 40 mg daily for 28 days and those allocated to PTX were given 400 mg three times per day for 28 days. The primary outcome measure was mortality at 28 days. Secondary outcome measures included mortality or liver transplant at 90 days and at 1 year. Rates of recidivism among survivors and the impact of recidivism on mortality were assessed. At 28 days, in arm A, 45 of 269 (16.7%) patients died; in arm B, 38 of 266 (14.3%) died; in arm C, 50 of 258 (19.4%) died; and in arm D, 35 of 260 (13.5%) died. For PTX, the odds ratio for 28-day mortality was 1.07 [95% confidence interval (CI) 0.77 to 1.40; p = 0.686)] and for prednisolone the odds ratio was 0.72 (95% CI 0.52 to 1.01; p = 0.056). In the logistic regression analysis, accounting for indices of disease severity and prognosis, the odds ratio for 28-day mortality in the prednisolone-treated group was 0.61 (95% CI 0.41 to 0.91; p = 0.015). At 90 days and 1 year there were no significant differences in mortality rates between the treatment groups. Serious infections occurred in 13% of patients treated with prednisolone compared with 7% of controls (p = 0.002). At the 90-day follow-up, 45% of patients reported being completely abstinent, 9% reported drinking within safety limits and 33% had an unknown level of alcohol consumption. At 1 year, 37% of patients reported being completely abstinent, 10% reported drinking within safety limits and 39% had an unknown level of alcohol consumption. Only 22% of patients had attended alcohol rehabilitation treatment at 90 days and 1 year. We conclude that prednisolone reduces the risk of mortality at 28 days, but this benefit is not sustained beyond 28 days. PTX had no impact on survival. Future research should focus on interventions to promote abstinence and on treatments that suppress the hepatic inflammation without increasing susceptibility to infection. This trial is registered as EudraCT 2009-013897-42 and Current Controlled Trials ISRCTN88782125. This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 19, No. 102. See the NIHR Journals Library website for further project information. The NIHR Clinical Research Network provided research nurse support and the Imperial College Biomedical Research Centre also provided funding.