Clinical risk classification is inaccurate in predicting relapse in adult patients with acute lymphoblastic leukemia, sometimes resulting in patients receiving inappropriate chemotherapy or stem cell transplantation (SCT). We studied minimal residual disease (MRD) as a predictive factor for recurrence and as a decisional tool for postconsolidation maintenance (in MRD(neg)) or SCT (in MRD(pos)). MRD was tested at weeks 10, 16, and 22 using real-time quantitative polymerase chain reaction with 1 or more sensitive probes. Only patients with t(9;22) or t(4;11) were immediately eligible for allogeneic SCT. Of 280 registered patients (236 in remission), 34 underwent an early SCT, 60 suffered from relapse or severe toxicity, and 142 were evaluable for MRD at the end of consolidation. Of these, 58 were MRD(neg), 54 MRD(pos), and 30 were not assessable. Five-year overall survival/disease-free survival rates were 0.75/0.72 in the MRD(neg) group compared with 0.33/0.14 in MRD(pos) (P = .001), regardless of the clinical risk class. MRD was the most significant risk factor for relapse (hazard ratio, 5.22). MRD results at weeks 16 to 22 correlated strongly with the earlier time point (P = .001) using a level of 10(-4) or higher to define persistent disease. MRD analysis during early postremission therapy improves risk definitions and bolsters risk-oriented strategies. ClinicalTrials.gov identifier: NCT00358072.