ObjectivesTo determine whether the use of recombinant human granulocyte colony-stimulating factor (G-CSF, filgrastim) reduces the mortality rate and the frequency rate of nosocomial infections in neutropenic patients requiring intensive care unit (ICU) admission.DesignRetrospective consecutive case series analysis.SettingMedical ICU of a teaching hospital.PatientsWe compared two groups of patients, according to whether or not they received G-CSF. In the ICU, 28 leukopenic patients received filgrastim (5 &mgr;g of body weight per day intravenously). In all these patients, G-CSF was continued until recovery from leukopenia, defined as a leukocyte count >1000/mm3. A total of 33 ICU leukopenic patients did not receive G-CSF. End points included leukocyte count, bone marrow recovery, frequency of ICU nosocomial infections (pneumonia, urinary tract, and catheter-related infections), and mortality rate.Measurements and Main ResultsThere were no differences in number of patients who recovered from leukopenia or in whom blood leukocyte count increased. Nosocomial infections occurred in the same percentage in both groups. The percentage of patients who died was identical in both groups. The percentage of patients with and without filgrastim therapy who recovered from leukopenia but died, was 86% and 78%, respectively.ConclusionIn the ICU, clinical outcome of neutropenic patients was not changed by G-CSF therapy. It is possible that G-CSF therapy may not be helpful in improving the ICU clinical outcome of neutropenic patients. Additional controlled studies designed to address this question are warranted.