The toxicity of autologous bone marrow transplantation (ABMT) is correlated to neutropenia. Although recombinant human granulocyte-macrophage colony-stimulating factor (rhu GM-CSF) seems to hold promise in accelerating neutrophil recovery, few analyses from randomized studies are presently available. Ninety-one patients with non-Hodgkin's lymphoma receiving high-dose ablative chemotherapy followed by ABMT with unpurged or purged marrow were included in a randomized, double-blind, placebo-controlled trial. Forty-four patients received 250 micrograms rhu GM-CSF (Escherichia coli)/m2 and 47 patients received placebo. Treatment was administered daily as continuous infusion from day of ABMT until the absolute neutrophil count (ANC) reached 0.5 x 10(9)/L for 7 days or until day 30, whichever was first. With rhu GM-CSF, 50% of the patients reached an ANC count greater than 0.5 x 10(9)/L at day 14 as opposed to day 21 with placebo (P less than .0001). Patients transplanted with marrow purged by mafosfamide also recovered earlier when treated with rhu GM-CSF (16 v 20.5 days, P = .013). The hospitalization duration was shorter in the rhu GM-CSF group (median, 23 v 28 days, P less than .05). No difference was observed in fever, number of infections, and antibiotic administration between the two groups. The major adverse event ascribed to rhu GM-CSF was a capillary leak syndrome in three patients graded as severe in two patients, moderate in one, and reversible in all three patients. In addition, one patient in the rhu GM-CSF group died suddenly with no explanation. In long term follow-up, the relapse rate was identical in both groups and there was no significant difference in the number of deaths at 1 year (12 with rhu GM-CSF v 9 with placebo), although deaths seemed to occur slightly earlier in the rhu GM-CSF group. We conclude that after ABMT with purged or unpurged marrow, rhu GM-CSF (E coli) significantly reduces neutropenia duration and hospitalization stay. A positive causative relation between the study drug and/or its mode of application with an increased toxicity as compared with GM-CSF from other sources and/or other modes of application cannot be deduced from the experiences in this study. Additional randomized trials would be necessary for an appropriate answer.