Temozolomide (TMZ) is currently being evaluated for the treatment of high-grade gliomas in children. Myelosuppression (the suppression of bone marrow activity) is the dose-limiting toxicity for TMZ in adults and children. Empirical methods (i.e. relations between the percent change in absolute neutrophil count (ANC) and the area under the plasma concentration curve (AUC) of TMZ or its active metabolite MTIC) showed poor results when attempting to describe myelosuppression from serial data derived during TMZ therapy in a Phase II study of children with high-grade glioma. Therefore, to improve our understanding of the myelosuppressive effects of TMZ and MTIC in children we developed a mechanistic mathematical model. The model describes the progression of neutrophils from their production in the bone marrow to their release in the plasma. Included in the model are the feedback effects of granulocyte colony stimulating factor (G-CSF), which stimulates neutrophil production when there is a decrease in circulating neutrophils. The model is fit to serial ANC measurements obtained after TMZ dosing and it is able to explain, among other things, the lag in ANC reduction following a dose of TMZ, the ANC nadir, and the 'rebound effect' observed where the ANC recovers to levels greater than that observed pre-TMZ dose. This model will be useful for the prospective design of clinical trials of TMZ in children with cancer.