To assess the clinical and hematopoietic effects of rhGM-CSF, a placebo-controlled double blind multicenter phase III study was undertaken in patients with non-Hodgkin's lymphoma receiving cytotoxic chemotherapy. Sixty-two patients who had granulocytopenia (<1×103/μl) after the first cycle of chemotherapy with cyclophosphamide, adriamycin, vincristine, and prednisolone were enrolled. After the second cycle of chemotherapy with the same regimen, patients randomly received either rhGM-CSF (125μg/m2/day) or placebo for 14 days (rhGM-CSF; 31 patients and placebo; 31 patients). Administration of rhGM-CSF induced a significant increase in granulocytes mainly with neutrophils, eosinophils and monocytes, but elevation of lymphocytes, latelets, and reticulocytes was not induced. Median days of granulocytes less than 1×103/μl in patients receiving rhGM-CSF were significantly shorter than in patients receiving placebo (p = O.001). Adverse reactions encountered with rhGM-CSF, and observed in 58% of the patients were never life-threatening and always rapidly reversible. They included fever, nausea and vomiting, diarrhea, skin eruption, and malaise. These results suggest that rhGM-CSF can be safely administered to prevent neutropenia after chemotherapy in patients with non-Hodgkin's lymphoma.