In the cooperative study COALL-80 151 children with acute lymphoblastic leukemia were treated according to the modified protocol BFM 79/81. The probability of continuous complete remission (CCR) for the total group is 74% after 6 years. In the subsequent study COALL-82 high-risk patients (initial white blood count greater than or equal to 25/nl or T-cell leukemia and acute undifferentiated leukemia) received additional high-dose methotrexate as fifth drug in the intensive phase and the combination VM-26/arabinosyl-cytosine in the reinduction phase. In all others patients (low-risk group) intermediate-dose methotrexate was substituted for the myelosuppressive agent cyclophosphamide in the intensive phase. Reinduction was no longer given in the low-risk group. In both risk groups cranial irradiation was postponed until after the intensive phase therapy. Radiotherapy was withheld for a group with minimal risk (white blood count less than 3/nl, liver/spleen less than 3 cm). The probability of CCR for the total group of 129 patients is 64% after 3.5 years. The comparative analysis between the studies COALL-82 and COALL-80 shows that low-risk patients have an equally high probability of continuous hematologic remission (91% vs. 85%) despite reduction of therapy in COALL-82. High-risk patients, however, had a significantly higher rate of bone marrow relapses; in this group the probability of continuous hematologic remission is 52% in COALL-82 vs. 73% in COALL-80. Also relapses in the central nervous system in irradiated patients were significantly more frequent in COALL-82 than in COALL-80 (12% vs. 4%) whereas the group of patients without radiotherapy has remained free of relapse. The possible influence of the modifications in therapy in study COALL-82 on the higher relapse rate is discussed.