In this prospective trial, a total of 74 children who were scheduled to undergo high-dose chemotherapy followed by autologous peripheral blood stem cell transplantation (PBSCT) were prospectively randomized at diagnosis to evaluate the effectiveness of exogenous granulocyte colony-stimulating factor (G-CSF) treatment in accelerating hematopoietic recovery after PBSCT. The diagnosis included acute lymphoblastic leukemia (ALL) (n = 27), neuroblastoma (n = 29), and miscellaneous solid tumors (n = 18). Eligibility criteria included (1) primary PBSCT, (2) chemotherapy-responsive disease, and (3) collected cell number >1 x 10(5) colony-forming unit-granulocyte-macrophage (CFU-GM)/kg and >1 x 10(6) CD34(+) cells/kg patient's body weight. After applying the above criteria, 11 patients were excluded due to disease progression before PBSCT (n = 6) or a low number of harvested cells (n = 5), leaving 63 patients for analysis; 32 patients in the treatment group (300 microg/m2 of G-CSF intravenously over 1 hour from day 1 of PBSCT) and 31 in the control group without treatment. Two distinct disease-oriented high-dose regimens without total body irradiation consisted of the MCVAC regimen using ranimustine (MCNU, 450 mg/m2), cytosine arabinoside (16 g/m2), etoposide (1.6 g/m2), and cyclophosphamide (100 mg/kg) for patients with ALL, and the Hi-MEC regimen using melphalan (180 mg/m2), etoposide (1.6 g/m2), and carboplatinum (1.6 g/m2) for those with solid tumors. Five patients (two in the treatment group and three in the control group) were subsequently removed due to protocol violations. All patients survived PBSCT. The median numbers of transfused mononuclear cells (MNC), CD34(+) cells, and CFU-GM were, respectively, 4.5 (range, 1 to 19) x 10(8)/kg, 8.0 (1.1 to 25) x 10(6)/kg, and 3.7 (1.2 to 23) x 10(5)/kg in the treatment group (n = 30) and 2.9 (0.8 to 21) x 10(8)/kg, 6.3 (1.1 to 34) x 10(6)/kg, and 5.5 (1.3 to 37) x 10(5)/kg, respectively, in the control group (n = 28), with no significant difference. After PBSCT, the time to achieve an absolute neutrophil count (ANC) of >0.5 x 10(9)/L in the treatment group was less than that in the control group (median, 11 v 12 days; the log-rank test, P =.046), although the last day of red blood cell (RBC) transfusion (day 11 v day 10) and the duration of febrile days (>38 degrees C) after PBSCT (4 v 4 days) were identical in both groups. However, platelet recovery to >20 x 10(9)/L was significantly longer in treatment group than control group (26 v 16 days; P =.009) and >50 x 10(9)/L tended to take longer in the treatment group (29 v 26 days; P =.126), with significantly more platelet transfusion-dependent days (27 v 13 days; t-test, P =.037). When patients were divided into two different disease cohorts, ALL patients showed no difference in engraftment kinetics between the G-CSF treatment and control groups, while differences were seen in those with solid tumors. We concluded that the marginal clinical benefit of 1 day earlier recovery of granulocytes could be offset by the delayed recovery of platelets. We recommend that the routine application of costly G-CSF therapy in children undergoing PBSCT should be seriously reconsidered.