Conventional therapy for childhood acute lymphoblastic leukemia (ALL) includes prednisone and oral 6-mercaptopurine. Prior observations suggested potential advantages for dexamethasone over prednisone and for intravenous (IV) over oral 6-mercaptopurine, which remain to be validated. We report the results of a randomized trial of more than 1000 subjects that examined the efficacy of dexamethasone and IV 6-mercaptopurine. Children with National Cancer Institute standard-risk ALL were randomly assigned in a 2 x 2 factorial design to receive dexamethasone (6 mg/m(2)/d) for 28 days in induction, plus taper, compared with prednisone (40 mg/m(2)/d). The second randomized assignment was for daily oral or weekly IV 6-mercaptopurine during consolidation. During maintenance, 5 days of the randomized steroid was given monthly, at the same dose, and all patients received daily oral 6-mercaptopurine. During delayed intensification, all patients received a dexamethasone dosage of 10 mg/m(2)/d for 21 days, with taper. Intrathecal (IT) methotrexate was the sole central nervous system-directed therapy. Patients randomly assigned to receive dexamethasone had a 6-year isolated central nervous system-relapse rate of 3.7% +/- 0.8%, compared with 7.1% +/- 1.1% for prednisone (P =.01). There was also a trend toward fewer isolated bone marrow relapses with dexamethasone. The 6-year event-free survival (EFS) was 85% +/- 2% for dexamethasone and 77% +/- 2% for prednisone (P =.002). EFS was similar with oral or IV 6-mercaptopurine; however, patients assigned to IV 6-mercaptopurine had decreased survival after relapse.