Primary end point of this trial was to reduce neutropenic infections during the treatment of aggressive NHL with CEOP/IMVP-Dexa (cyclophosphamide, epirubicin, vincristine, prednisolone ifosfamide, methotrexate, VP-16, and dexamethasone). Further, we studied the influence of filgrastim on dose intensity of CEOP/IMVP-Dexa, on the rate of complete remissions, on the time to relapse, and on survival. Eighty-five patients with untreated large-cell NHL were randomized to one of two treatment arms; 74 patients were eligible. Thirty-eight patients in arm 1 were treated with CEOP/IMVP-Dexa chemotherapy and filgrastim, 36 in arm 2 with CEOP/IMVP-Dexa chemotherapy alone. In arm 1 filgrastim was self-injected by the patients at 5 micrograms/kg body wt. s.c. daily, except on the days when cytotoxic drugs were given. During treatment we did weekly complete blood counts. Median leukocyte counts were 10.91 x 10(9)/l and 5.46 x 10(9)/l in arm 1 and 2, respectively (p = 10(-6)). Median neutrophil counts were 7.7 x 10(9)/l in arm 1 and 2.72 x 10(9)/l in arm 2 (p < 10(-6)). Median neutrophil nadirs were 0.199 x 10(9)/l and 0.213 x 10(9)/l in arm 1 and 2, respectively (p = 0.09). Mean platelet nadirs were 95 and 152 x 10(9)/l (p = 0.000004) and mean hemoglobin nadirs 83.95 g/l and 92.78 g/l (p = 0.00558) in arm 1 and 2, respectively. Dose intensity of CEOP/IMVP-Dexa was 82.3% and 76.2% in arm 1 and 2, respectively (p = 0.041). Forty-two percent and 58% of patients experienced a febrile neutropenia in arm 1 and 2, respectively (not significant, NS). Median time to first neutropenic infection was in treatment week 11 and 6 in arm 1 and 2, respectively (NS). There was no significant difference in rate, duration, and kind of infection, duration of hospitalization, or antibiotic treatment. Seven toxic deaths occurred, all due to neutropenic infection, 6 and 1 in arm 1 and 2, respectively (p = 0.0732). Four of the six patients, who died of infection in arm 1 were older than 60 years. Complete remission rate was 83% and 66.7% in arm 1 and 2, respectively (NS). After a median observation time of 3 years there was no difference in time to relapse or survival. Filgrastim increases leukocyte and neutrophil counts and dose intensity, if used with CEOP/IMVP-Dexa chemotherapy in high-grade lymphomas. There was no significant effect on febrile neutropenia or infections. The more frequent fatal neutropenic infection rate in the filgrastim arm was not statistically significant. It is most appropriate to explain it by the patient's age in combination with the high dose intensity. The small increase in dose intensity had no effect on survival but probably decreased hemoglobin levels and platelet counts in arm 1. We were unable to show a benefit for filgrastim in combination with CEOP/IMVP-Dexa.