Four thousand, four hundred and forty eligible children of up to 18 years of age were treated in four consecutive trials between 1981 and 1995 with the treatment protocols of the Berlin-Frankfurt-Münster (BFM) study group for childhood acute lymphoblastic leukemia (ALL). The probability for event-free survival (pEFS) at 8 years improved from 65.8% in study ALL-BFM 81 to 75.9% in study ALL-BFM 90. The cumulative incidence of recurrences with CNS involvement was 10.1% and 9.3% in studies ALL-BFM 81 and 83, but was reduced to less than 5% in study ALL-BFM 90 (for isolated CNS relapses from 5.3% in study ALL-BFM 81 to 1.1% in study ALL-BFM 90). Four major findings were derived from this series of trials performed by 37 to 96 centers in Germany, Austria, and Switzerland: (1) Reintensification is a crucial part of treatment, even in low risk patients; (2) presymptomatic cranial radiotherapy can be safely reduced to 12 Gy, or even be eliminated if it is replaced by early intensive systemic and intrathecal methotrexate applied; (3) maintenance therapy given a total of 24 months from diagnosis provides a lower rate of systemic relapses than treatment for 18 months; (4) inadequate response to an initial 7-day prednisone window (combined with one intrathecal injection of methotrexate on day 1) defines about 10% of the patients with a very high risk of relapse. For patients with adequate early response (90% of all) an 8-year pEFS of 80% has been achieved in the most recent trial ALL-BFM 90. While it has proven so far to be impossible to improve the outcome for the small group of high risk patients, the number of recurrences could be effectively reduced for the large group of patients responding adequately to the prednisone in vivo sensitivity test. Apart from inadequate prednisone response, patients with hyperleukocytosis, age <1 year, or the presence of the Philadelphia-chromosome (Ph+ ALL) are at a particularly high risk of failure.