Background: Supportive-care use of granulocyte colony-stimulating factor (G-CSF) in pediatric acute myeloid leukemia (AML) remains controversial due to a theoretical increased risk of relapse and limited impact on neutropenic complications. We describe the use of G-CSF in patients treated according to NOPHO-AML 2004 and DB AML-01 and investigated associations with relapse. Procedure: Patients diagnosed with de novo AML completing the first week of therapy and not treated with hematopoietic stem cell transplantation in the first complete remission were included (n = 367). Information on G-CSF treatment after each course (yes/no) was registered prospectively in the study database and detailed information was gathered retrospectively from each center. Descriptive statistics were used to describe G-CSF use and Cox regression to assess the association between G-CSF and risk of relapse. Results: G-CSF as supportive care was given to 128 (35%) patients after 268 (39%) courses, with a large variation between centers (0-93%). The use decreased with time—the country-adjusted odds ratio was 0.8/diagnostic year (95% confidence interval [CI] 0.7-0.9). The median daily dose was 5 μg/kg (range 3-12 μg/kg) and the median cumulative dose was 75 μg/kg (range 7-1460 μg/kg). Filgrastim was used in 82% of G-CSF administrations and infection was the indication in 44% of G-CSF administrations. G-CSF was associated with increased risk of relapse—the adjusted hazard ratio was 1.5 (95% CI 1.1-2.2). Conclusions: G-CSF as supportive care was used in a third of patients, and use decreased with time. Our results indicate that the use of G-CSF may be associated with an increased risk of relapse.