The toxicity, pharmacokinetics, and hematologic effects of granulocyte-macrophage colony-stimulating (GM-CSF) were studied in a phase I/II trial of 16 patients with myelodysplastic syndrome (MDS). The GM-CSF was administered subcutaneously (SC) daily so as to achieve prolonged blood levels and to establish an outpatient treatment regimen. Four dose levels were administered for ten days: 0.3 microgram/kg/d (three patients), 1.0 microgram/kg/d (three), 3.0 micrograms/kg/d (four), and 10.0 micrograms/kg/d (six). The most common toxicities were fever and a flu-like syndrome, which were dose-dependent. The maximum-tolerated dose was 10.0 micrograms/kg/d, which induced severe rigors (two patients), fever greater than 40 degrees C (one), severe bronchospasm (one), and WBC 60,000 (one). In one patient, refractory anemia with excess blasts in transformation (RAEB-T) progressed to acute nonlymphocytic leukemia after two doses of GM-CSF, and the patient died of leukemia that did not respond to chemotherapy. After doses of 3.0 and 10.0 micrograms/kg, serum GM-CSF levels peaked at 3.8 to 6.3 hours, and persisted for 14 and 24 hours, respectively. Circulating granulocytes (neutrophils and bands) increased in a dose-dependent manner, as 11 of 13 patients who received greater than or equal to 1.0 microgram/kg/d responded with a two- to 194-fold increase. Although the neutrophils usually returned to pretreatment levels shortly after stopping GM-CSF, two patients continue to exhibit an elevation of neutrophils for 6 months. Dose-related increases in circulating monocytes and eosinophils were also noted. Transient increases in platelet and reticulocyte counts were observed in two and three patients, respectively. Five of the 16 patients later received maintenance GM-CSF at 3 micrograms/kg/d for 2 to 9 weeks. All showed a dramatic increase in neutrophils after 2 weeks. Thereafter, despite continued therapy, the neutrophil count in four patients declined markedly. In conclusion, GM-CSF is well tolerated by the SC route and induces striking, but usually temporary, improvement in the neutropenia of MDS. Larger prospective phase III trials will determine the duration of hematologic responses and the impact on infection, morbidity, and mortality.