PURPOSE Measurable residual disease (MRD) by using flow cytometry after induction therapy is strongly prognostic in pediatric AML, and hematopoietic stem-cell transplant (hSCT) may counteract a poor response. We designed a phase III study with intensified response-guided induction and MRD-based risk stratification and treated poor induction response with hSCT. The efficacy of liposomal daunorubicin (DNX) in induction was compared with mitoxantrone. METHODS The study planned to randomly assign 300 patients, but the production of DNX ceased in 2017. One hundred ninety-four patients were randomly assigned to mitoxantrone or experimental DNX in induction 1. Ninety-three non–randomly assigned patients served as an observation cohort. Primary end point was fraction of patients with MRD <0.1% on day 22 after induction 1. Patients with MRD ≥15% after induction 1 or ≥0.1% after induction 2 or FLT3-ITD with NPM1 wildtype were stratified to high-risk therapy, including hSCT. RESULTS Outcome for all 287 children was good with 5-year event-free survival (EFS_5y) 66.7% (CI, 61.4 to 72.4) and 5-year overall survival (OS_5y) 79.6% (CI, 75.0 to 84.4). Overall, 75% were stratified to standard-risk and 19% to high-risk. There was no difference in the proportion of patients with MRD <0.1% on day 22 after induction 1 (34% mitoxantrone, etoposide, araC [MEC], 30% DNX, P = .65), but the proportion increased to 61% for MEC versus 47% for DNX ( P = .061) at the last evaluation before induction 2. EFS_5y was significantly lower, 56.6% (CI, 46.7 to 66.5) versus 71.9% (CI, 63.0 to 80.9), and cumulative incidence of relapse (CIR) was higher, 35.1% (CI, 25.7 to 44.7) versus 18.8% (CI, 11.6 to 27.2) for DNX. The inferior outcome for DNX was only in standard-risk patients with EFS_5y 55.3% (CI, 45.1 to 67.7) versus 79.9% (CI, 71.1 to 89.9), CIR 39.5% (CI, 28.4 to 50.3) versus 18.7% (CI, 10.5 to 28.7), and OS_5y 76.2% (CI, 67.2 to 86.4) versus 88.6% (CI, 81.4 to 96.3). As-treated analyses, including the observation cohort, supported these results. For all high-risk patients, 85% received hSCT, and EFS_5y was 77.7 (CI, 67.3 to 89.7) and OS_5y was 83.0 (CI, 73.5 to 93.8). CONCLUSION The intensification of induction therapy with risk stratification on the basis of response to induction and hSCT for high-risk patients led to improved outcomes. Mitoxantrone had a superior anti-leukemic effect than liposomal daunorubicin.