BACKGROUND In this retrospective, uncontrolled, observational study, the effect of granulocyte colony-stimulating factor (G-CSF)-stimulated granulocyte transfusions (GTX) in neutropenic paediatric patients with sepsis was evaluated. METHODS Granulocytes were collected from unrelated, ABO group-matched and cytomegalic-antibody compatible donors. For neutrophil mobilization, donors received a single subcutaneous dose of glycosylated G-CSF (Lenograstim, Chugai Pharma, Japan) plus oral dexamethasone (8 mg). In total, 168 (range 1-19 per patient) GTX were transfused in 32 children with a median age of 7.4 (0.25 to 16) years. RESULTS The underlying diseases comprised predominantly haematooncological malignancies (31 children). In 15 of 32 patients, neutropenia was related to allogeneic stem cell transplantation. All children suffered from sepsis based on international criteria (fever, tachycardia, respiratory rate >2 SD above normal in the context of a suspected or proven infection). In ten children bacteria were isolated, in six children a fungal infection was diagnosed and four sepsis episodes were caused by viral infections. GTX contained a median neutrophil number of 6.3 (range 1.9-13.9)x10(10) per transfusion and obtained a sustained haematological response after GTX. Nineteen out of 32 children survived the neutropenic sepsis, particularly nine out of 11 patients with bacterial sepsis. CONCLUSIONS In contrast to the non-survivors, we observed a significant decrease in the C-reactive protein levels shortly after initiation of the GTX treatment in the surviving patients. A clear-cut benefit of GTX for children with neutropenic sepsis cannot be concluded from these data, but in children with (severe) bacterial sepsis refractory to antibiotic treatment, GTX were feasible, safe and could reduce mortality rates in this subgroup of patients.