In the early stages of the development of granulocytic colony-stimulating factors (G-CSF and GM-CSF) in oncology and hematology, myeloid malignancies were considered to be a contraindication to their use. In fact, myeloid leukemic cells bear specific receptors for G-CSF and GM-CSF and these CSFs induce an in vitro proliferation in primary blast cells of most patients with acute myeloid leukemia (AML). In addition, autocrine or paracrine loops of stimulation have been demonstrated in some cases. Despite these theoretical risks of blast proliferation, G-CSF and GM-CSF have been extensively tested in patients with AML or myelodysplastic syndromes. Major objectives were the correction of acquired or chemotherapy-induced neutropenia, but also the reinforcement of the antileukemic efficacy of cytotoxic agents. Recently, G-CSF has also been used to mobilize hematopoietic progenitors in the peripheral blood. Major results of several double-blind clinical trials are the demonstration of the safety of CSF administration in these patients, since no risk of in vivo blast cell regrowth has been observed, and their efficacy to shorten the duration of chemotherapy-induced neutropenia. However, no significant reduction in the treatment-related mortality and no survival improvement were afforded by the use of these CSFs. From another point of view, the search for AML-specific CSF-receptor or CSF-receptor associated molecule abnormalities represents a new promising area to try to understand the mechanisms of leukemogenesis.