Purpose: This multicenter pilot study was aimed to evaluate the feasibility of new treatment protocol based on risk stratification for newly diagnosed pediatric de novo acute myeloid leukemia (AML). Methods: Acute promyelocytic leukemia and AML of Down syndrome were not eligible. Risk features were defined as follows: 1) low risk features (LRF): inv(16)/t(16;16), t(8;21) without c-kit mutation, or with normal karyotype in the presence of NPM1 or CEBPA mutation without FLT3/ITD mutation; 2) high risk features (HRF): -5, 5q-, -7, 3q abnormalities, t(8;16), t(6;9), t(16;21), t(6;11), t(10;11), complex karyotype, or FLT3/ITD mutation; 3) standard risk features (SRF): all the others. A new chemotherapy regimen was developed which consisted of 2 induction and 4 consolidation cycles. Treatment responses were determined according to the marrow response at the end of each induction as good response (CR-CR), delayed response-1 ( PR [blast 5~15%]-CR), delayed response-2 (NR [blast＞15%]-CR), refractory ( PR/NR-PR/NR), and early relapse. Patients were allocated into either favorable (FG), intermediate (IG), or poor ( PG) prognostic group considering both risk features and treatment response. Patients in FG were not candidates for transplantation. Results: Thirty-four patients were enrolled. One patient (2.9%) had CNS leukemia and 8 patients (23.5%) had chloroma. There were 8 patients (23.5%) with LRF, 16 (47.1%) with SRF, and 10 (29.4%) with HRF. The overall CR rate following 2 cycles of induction was 96.7%. Among 30 patients who completed 2 cycles of induction, 8 patients (26.7%) belong to FG, 13 (43.3%) to IG, and 9 (30%) to PG. Neutropenic fever and blood stream infection developed in 64~97% and 12~29% of patients during each chemotherapy cycle, while there was no proven invasive fungal infection. Eleven patients underwent allo-HSCT. With a median follow-up of 10 mo, the estimated 1-y overall survival and event-free survival were 100% and 89%, respectively. One patient relapsed after completion of chemotherapy and the other one relapsed after transplantation. There was no treatment-related mortality. Conclusion: Although the follow-up duration is short, our new treatment protocol based on risk stratification seems safe and effective. We suggest a nationwide prospective multicenter trial to further test the feasibility of this strategy for pediatric de novo AML.