This controlled study of children with ALL was designed to test the efficacy and toxicity of one-, two-, three- and four-drug therapy during remission and whether more aggressive therapy in the first eight weeks prolongs remission in patients with features associated with a particularly poor prognosis. After inducing remission with prednisone, vincristine and asparaginase, patients received cranial irradiation and IT methotrexate and were randomized to receive: 1--methotrexate alone; 2--methotrexate plus mercaptopurine; 3--same as in group 2 plus cyclophosphamide; and 4--same as in group 3 plus arabinosyl cytosine. Patients with CNS leukemia at diagnosis received IT methotrexate weekly during the induction period and a higher dose of CNS irradiation. Patients with anterior mediastinal enlargement at diagnosis received radiotherapy to the mass during the induction period. Patients who failed to attain bone marrow remission after four weeks of therapy were given daunorubicin and prednisone for 2--4 additional weeks. Of the 282 patients entering this study between January 1972 and November 1975, 268 (95%) attained complete remission and 228 (85%) were randomized to receive continuation chemotherapy with 1, 2, 3 or 4 drugs. In Group 1 (methotrexate alone), 14 of 20 patients relapsed and 9 developed leukoencephalopathy without antecedent CNS leukemia apparently due to higher doses of intravenous methotrexate; in Groups 2, 3 and 4 the results were equivalent, but without leukoencephalopathy in initial CR. The addition of cyclophosphamide and arabinosyl cytosine increased toxicity and complications without demonstrably increasing the leukemocidal effect. In the 40 patients given additional early therapy, the modalties employed in this study did not prolong remission.