Background. The efficacy in pediatric acute myeloid leukemia (AML) of single-agent methotrexate (MTX) at a higher dose than previously applied, 1,000 mg/m2, given as a theoretically beneficial 36-hr continuous infusion, is unknown, but may be beneficial based on preclinical data. Procedure. We performed a therapeutic window study in children with first relapsed AML treated in four different countries. Results. Based on a comparison between the percentage of leukemic blasts in the bone marrow shortly before and 7-10 days after the MTX infusion, none of the first cohort of nine patients showed a good response, defined as a reduction of blasts of at least 50%. Therefore, the study was closed, concluding that the probability of a good response in this patient-group was most likely to be less than 30%. By that time, another four patients had been enrolled, of which one patient with a late relapsed AML FAB type M7 showed a good response. Toxicity of MTX was limited and tolerable. Conclusions. This study shows that single-agent MTX in the applied regimen in pediatric relapsed AML has limited efficacy. However, it also demonstrates the feasibility of an international and therapeutic window phase II study in pediatric relapsed AML.