Fifty-one children between 26 and 214 months of age (median, 100 months) with acute lymphoblastic leukemia (ALL) were grafted in second remission from HLA-identical sibling donors (except for two patients who were grafted with a marrow with 1 antigen-mismatch). Initial treatment and relapse therapy were similar in all patients according to the BFM- and CoALL-protocols (front line: 38 patients according to BFM-protocols and 13 patients according to CoALL-protocols; relapse: 12 patients in study ALL-REZ-BFM 83, 17 in ALL-REZ-BFM 85, 20 in ALL-REZ-BFM 87, and two in ALL-REZ-BFM 90). The conditioning regimens were different, consisting of cyclophosphamide (CY) total body irradiation (TBI) plus (n = 27), VP-16-TBI (n = 23), and CY-TBI and ARA-C (n = 1). Three patients had a second graft after conditioning with CY-TBI for the first transplantation. The second ablative regimen consisted of CY plus VP-16 in the first patient and CY plus busulfan in the two other patients, one of whom relapsed again. All patients but three had bone marrow (BM), either isolated or combined, relapses. Twenty-nine of the patients are in continuous complete remission (CCR), ranging from 1 to 67 months after transplantation with a median time of 30 months. One patient was lost to follow-up in continuous remission. Nine patients died from treatment-related complications (infections and graft-versus-host disease) and 12 patients suffered a leukemic relapse; three of them received a second graft and two are in CCR. Kaplan-Meier analysis yields an event-free survival (EFS) of 0.52 +/- 0.08. The probability of a 7-year relapse-free interval (RFI) is 0.68 +/- 0.08. EFS for patients with late relapses is 0.47 +/- 0.12 and for patients with early relapses 0.56 +/- 0.1. The RFI for patients with late relapses is 0.65 +/- 0.12 and for patients with early relapses 0.69 +/- 0.11. There is a nonsignificant trend towards superior results for patients grafted after conditioning with VP-16 plus TBI. When all patients who are not in CCR at day +125 (which is the median interval between relapse diagnosis and BM transplantation [BMT]) are excluded from the chemotherapy results, there is no significant difference between the results of BMT and chemotherapy for late relapses. On the other hand, there is a significant advantage between chemotherapy and BMT for early relapses over chemotherapy (P less than or equal to .01).