We conducted a clinical trial and report the long-term outcome of 773 children with acute lymphoblastic leukemia upon risk-adapted therapy accrued in trial CoALL 07-03 (from the Cooperative Study Group for Childhood Acute Lymphoblastic Leukemia). In a 2-step stratification, patients were allocated to receive either low- or high-risk treatment, based on initial white blood cell count, age, and immunophenotype. A second stratification was performed according to the results of in vitro pharmacosensitivity toward prednisolone, vincristine, and asparaginase (PVA score) and in vivo response after induction therapy (minimal residual disease [MRD]). Therapy was reduced for both risk groups in patients with a low PVA score or negative MRD result, and intensified in patients with a high PVA score. Overall outcome improved significantly compared with the predecessor CoALL 06-97 trial, with identical therapy backbone despite treatment reduction in 15.8% of patients (10-year probability of event-free survival, 83.5% vs 73.9%; overall survival, 90.7% vs 83.8%). Outcome for patients in the reduced treatment arms was superior to that of patients in the standard arms, associated with a profound reduction in frequency and severity of infectious complications. Importantly, we observed a lack of correlation between in vitro and in vivo drug response, as well as a lower predictive value of in vitro drug testing, reflecting an intrinsic limitation of this methodology that prevents its use for treatment stratification in future trials. In conclusion, it might be possible to reduce chemotherapy in children with acute lymphoblastic leukemia selected by stringent in vivo measurement of MRD without jeopardizing overall outcome.