Abstract Mitoxantrone (Novantrone, American Cyanamid Company; NO) and high-dose cytarabine (Ara-C; AC) have each been shown to be active in non-Hodgkin's lymphomas (NHL) in various studies. The studies reported here are sequential. The first study (NOAC I) combined high-dose cytarabine (3 g/m 2/12 h as a 3 h infusion on day 1) with mitoxantrone (10 mg/m 2/d on days 2 and 3). Of 31 patients with relapsed and refractory NHL, 7 achieved complete remission (CR) and 7, partial remission (PR). Myelosuppression was the major toxicity of this regimen. In the second study (NOAC II), the dosage of cytarabine was escalated to 3 g/m 2/12 h on days 1 and 2 (4 doses) while mitoxantrone remained 10 mg/m 2/d on days 2 and 3. The effects of recombinant human (rh) granulocytemacrophage colony-stimulating factor (GM-CSF) were simultaneously studied. Twenty-three patients from five centers were treated with NOAC plus rhGM-CSF while 14 patients from four centers received NOAC II alone. A CR was achieved in 9 of 23 patients who received the additional rhGM-CSF and in 2 of 14 patients treated with NOAC alone. With rhGM-CSF, the median duration of severe neutropenia (<0.5/nL) after chemotherapy was 8 days versus a median of 13 days without rhGM-CSF, while the duration of severe thrombocytopenia (<20/nL) was not significantly different. The rates of infection and mucositis were 25% and 17%, respectively, with rhGM-CSF compared to 53% and 60% without rhGM-CSF. Thus, this last nonrandomized pilot study indicates that administration of rhGM-CSF reduces the duration of chemotherapy-induced cytopenia and the rate of mucositis. This growth factor does not appear to result in stimulation of lymphoma cells. At present, a controlled randomized trial is being conducted using NOAC II with rhGM-CSF or placebo to establish the definitive role of this growth factor in the treatment of NHL.