Recombinant human interleukin 3 (rhIL-3) has been suggested to be a useful agent for the treatment of chemotherapy-induced thrombocytopenia. For evaluation of this possibility, rhIL-3 was given subcutaneously for 10 days to patients with small-cell lung cancer (SCLC). Chemotherapy consisted of carboplatin (CBDCA) given at 400 mg/m2 to previously untreated patients or at 350 mg/m2 to previously treated patients on day 1 and etoposide (VP-16) given at 100 mg/m2 on days 1-3 every 4 weeks. If the platelet count nadir was < 75,000/microliters in the control cycle of chemotherapy, patients were randomly assigned for the next cycle to rhIL-3 given at 5 or 10 micrograms/kg per day on days 4-13. A total of 41 patients (32 previously untreated patients and 9 previously treated patients) were enrolled in the study. The platelet count nadir in the cycles including rhIL-3 was significantly higher at both dose levels (P < 0.01) than in the control cycle. The duration of thrombocytopenia (< 75,000/microliters) and the mean time from the 1st day of chemotherapy to thrombocyte recovery (> 100,000/microliters) in the rhIL-3 cycle were significantly shorter than those in the control cycle (P < 0.01). The neutrophil count nadir and the duration of neutropenia (<1,000/microliters) were also significantly improved in the rhIL-3 cycle (P < 0.05). The major side effects were fever (80.5%), headache (24.3%), and fatigue (14.6%). All side effects were tolerable and of less than grade II. There was no difference in the efficacy of the two dose levels, but the 5-micrograms/kg dose appeared to be better tolerated than the 10-micrograms/ kg dose. We conclude that rhIL-3 administration following chemotherapy consisting of CBDCA and VP-16 reduces the incidence and severity of chemotherapy-induced thrombocytopenia and neutropenia with an acceptable adverse-events profile.