Ninety-five patients undergoing an allogeneic bone marrow transplant (BMT) and developing acute graft-versus-host disease (aGvHD) were randomized to receive low-dose intravenous 6-methylprednisolone (6MPred; 2 mg/kg /d; n = 47) or high-dose 6MPred (10 mg/kg/d; n = 48) for 5 days, with subsequent tapering doses. On day 5 patients not responding or progressing on low-dose 6MPred could be switched to high-dose 6MPred. All patients, aged 1 to 55 years, were recipients of unmanipulated BMT from HLA identical sibling donors. Patients were stratified at randomization for age (v10 mg/kg) were as follows: response of aGvHD 68% versus 71% (P= .9), evolution to aGvHD grade III-IV 17% versus 20% (P= .6), CMV infections 55% versus 60% (P = .7), 3-year actuarial TRM 28% versus 32% (P = .7), relapse 17% versus 7% (P = .1). The actuarial survival at 3 years was 63% versus 62% (P = .9) with a median follow up of 580 and 778 days. On day 5 of therapy, 26 patients assigned to low-dose (2 mg/kg) 6MPred were switched to a higher dose of 6MPred because of no response or progression. Their actuarial TRM was 46%, which is significantly higher than TRM of patients who responded on 2 mg/kg and continued with tapering doses (TRM = 16%, P = .007). In conclusion, early treatment of acute GvHD with 6MPred 10 mg/kg/d does not improve the response rate as compared with 2 mg/kg/d, nor does it prevent evolution to aGvHD grade III-IV. CMV infections, TRM, and survival were also comparable. A group of patients at high risk of TRM can be identified after 5 days of treatment with 6MPred 2 mg/kg and could be eligible for alternative forms of therapy.