Although polymorphonuclear leukocytes (PMNs) are powerfully anti-Aspergillus, transfusion therapy remains controversial, with conflicting results, and experimental support has been lacking. We devised a pulmonary infection model in neutropenic BALB/c mice, used an antibacterial regimen to prevent confounding sepsis, and optimized PMN induction, purifications, and dose. Mice were given 150 mg/kg cyclophosphamide every 4 days and a gentamicin-vancomycin-clindamycin-imipenem regimen daily beginning 4 days before intranasal challenge with 5 × 10(5) Aspergillus conidia. This regimen produced leukopenia (~10% of normal white blood cell [WBC] count; ≤ 10% PMNs) for 10 days, without bacterial superinfection. PMN donors given 100 μg/kg recombinant murine granulocyte colony-stimulating factor (G-CSF) for 10 days yielded 11 × 10(7) to 13.6 × 10(7) WBC/ml (81 to 87% PMNs). Infected mice were given PMN transfusions intravenously. In 2 experiments with up to 70% mortality of neutropenic controls, transfusion of 10(7) PMNs 1 and 4 days after challenge had negligible effects on peripheral WBC counts but improved survival (P = 0.007, 0.02), decreased lung CFU (P = 0.03, 0.005), and cleared infection in 28 to 50% of survivors. Transfusion of 5 × 10(6) PMNs showed partial protection. Transfusions given every other day did not improve protection. Our present results provide an experimental basis for enthusiasm for PMN transfusions in the therapy of aspergillosis in humans.