To determine whether recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF) can offset the myelosuppressive effects of intensive chemotherapy, we carried out a double-blind placebo-controlled trial in which 40 patients with acute lymphoblastic leukemia (ALL) were randomized into two groups of 20 each. One group received rhGM-CSF (5.5 micrograms/kg SC) coadministered with chemotherapy and the other, placebo coadministered with chemotherapy from day 5 to day 11 and from day 19 to day 25 of the 28-day intensification phase of our institutional high-risk protocol for childhood ALL. The results indicate that, at the dose and schedule used, rhGM-CSF did not prevent neutropenia or shorten the number of days required to complete this phase of therapy. In addition, the treated and placebo groups showed no significant difference in absolute neutrophil counts, number of days with neutropenia, number of days with fever, number of days spent in hospital, or number of days on antibiotics during the 28-day study period. There was also no difference between the two groups in the number, type, or severity of infectious episodes. Two of 20 patients in the treatment group have relapsed, whereas none of the patients in the placebo group has yet relapsed (follow-up: 3-37 months), but these events were not statistically significant. We conclude that treatment with rhGM-CSF at the dose and schedule employed is not clinically beneficial.