BACKGROUND To reduce morbidity and mortality adjuvant cytokine therapy was administered to septic neonates with variable results. The objective of this case series was to compare the effectiveness of recombinant human granulocyte-macrophage colony-stimulating factor (rhuGM-CSF) and recombinant granulocyte colony-stimulating factor (rG-CSF) with that of placebo in correcting neutropenia induced by sepsis. METHODS Symptomatic, septic premature neonates with or without a positive blood culture were eligible. Twenty-eight patients were randomized: 10 received rG-CSF (5 microg/kg/dose i.v. twice a day); 10 received rhuGM-CSF (4 microg/kg/dose i.v. twice a day) and 8 received placebo for a maximum of 7 days, or until an absolute neutrophil count (ANC) of 10,000 cells/mm was reached. RESULTS A significant increase in the ANC above the baseline was present on Day 2 in the rG-CSF group (P = 0.015) and on Day 5 in the rhuGM-CSF (P = 0.002) and placebo (P = 0.027) groups. The ANC of the rG-CSF group was significantly above that in the rhuGM-CSF and placebo groups on Day 7 (P = 0.03). Mortality and neonatal intensive care unit morbidity was not significantly different between the groups. CONCLUSIONS The neutrophil count in the rG-CSF-treated group increased significantly faster than that in the placebo or rhuGM-CSF group.