The present trial was designed to test the effects of G-CSF on the duration of the second phase of induction chemotherapy in children with newly diagnosed acute lymphoblastic leukemia (ALL). A total of 32 patients were assigned randomly to a group that received (14 patients; group A) or a group that did not receive (18 patients; group B) G-CSF (10 g/kg/day subcutaneously and daily) throughout of the second phase of induction therapy. One of 14 (7.1%) patients in group A and 2 of 18 (11.1%) patients in group B completed the course of chemotherapy within the planned time. The median length of this phase was 37 days (range, 29 to 65; mean, 40; SD, 8.6) for patients in group A and 36 days (range, from 29 to 55; mean, 38; SD, 7.4) for those in group B, and the difference was not statistically significant. The number of days during which patients had granulocyte counts of less than 2 x 10(9)/l, the number of febrile episodes of unknown origin, the number of bacterial and fungal infections and the number of days of hospitalization did not differ in a statistically significant manner between the two groups. Our data suggest that G-CSF supportive therapy may be unnecessary in children with neutropenia of short duration, for whom the risk of infection is low.