To determine the efficacy of combined G-CSF and highly active antiretroviral treatment (HAART), a randomized, double blind, placebo controlled study was conducted. Treatment naive human immunodeficiency virus (HIV) infected patients were randomized to receive either placebo or G-CSF (0.3 mg/ml, 3 times a week) for 12 weeks and HAART simultaneously. The trial was terminated prematurely after interim analysis performed because of a case of severe encephalopathia in the G-CSF group. At that point 11 HIV infected patients with a CD4+ T cell count < 350/mm3 had been randomized to the G-CSF group (n = 6) or placebo group (n = 5). In both groups plasma HIV RNA decreased significantly in response to HAART. However, plasma HIV RNA changed significantly different between the two groups with the decrease being less pronounced in the G-CSF group (P = 0.02). The concentrations of CD4+ memory T cells and CD8+ naive and memory T cells increased in response to HAART, and there was a trend towards more pronounced increases in several T-cell subpopulations in the G-CSF group. The CD56+ NK cells increased significantly more in the G-CSF group compared with placebo (P = 0. 000). All patients in the G-CSF group reported bone pain. The present data do not support simultaneous administration of G-CSF with initiation of HAART in treatment naive HIV infected patients.