The aim of the study was to analyze the impact of minimal residual disease (MRD) after reinduction therapy on the outcome of children with relapsed 'high-risk' acute lymphoblastic leukemia (ALL). Sixty patients with isolated or combined marrow relapse were studied. All patients belonged to the S3 or S4 groups, as defined by the Berlin-Frankfurt-Münster stratification for relapsed ALL. MRD was studied by real-time quantitative PCR after the first, second and third chemotherapy course (time points 1 (TP1), 2 (TP2) and 3 (TP3), respectively). MRD results, not used for treatment refinement, were categorized as negative (NEG MRD), positive not-quantifiable (POS-NQ MRD) when MRD level was below quantitative range (a level <10(-4)) or positive within quantitative range (POS MRD) when MRD level was >or=10(-4). With a median observation time of 15 months, overall 3-year event-free survival (EFS) was 27%. The 3-year EFS was 73, 45 and 19% for patients with NEG-MRD, POS NQ-MRD and POS-MRD at TP1, respectively (P<0.05). The prognostic predictive value of MRD was statistically confirmed in multivariate analysis. MRD quantitation early and efficiently differentiates patients who benefit from conventional treatment, including allogeneic hematopoietic stem cell transplantation, from those needing innovative, experimental therapies.