In acute lymphoblastic leukemia (ALL), treatment with granulocyte colony stimulating factor (G-CSF) during remission induction shortens granulocytopenia and may decrease morbidity due to infections. However, the optimal timing of G-CSF administration after chemotherapy is not known. In a prospective randomized multi-center study, adult ALL patients were treated with high-dose ARA-C [HDAC, 3 g/m(2) bid (1 g/m(2) bid for T-ALL) days 1-4] and mitoxantrone (MI 10 mg/m(2) days 3-5). They were randomized to receive recombinant human G-CSF (Lenograstim) 263 micro g/day SC starting either from day 12 (Group 1) or day 17 (Group 2). Fifty-five patients (41 male, 14 female) with a median age of 34 years (range: 18-55 years) were enrolled into the study; 50 patients were evaluable. The median duration of neutropenia <500/ micro l after HDAC/MI was 12 days (range: 7-22 days) in the early G-CSF Group 1 and also 12 days (range: 4-22 days) in the late G-CSF Group 2; this was shorter than in the historical control group (15 days, range: 4-43 days, n=46) where the patients received identical cytotoxic treatment without G-CSF. Seventeen infections were observed in 14 patients in Group 1 (47%) and 13 infections in 10 patients in Group 2 (50%) compared to 27 infections in 49 patients of the historical control (54%). In Group 1, the patients received a median of 11 injections with G-CSF (range: 7-22) compared to 7 injections (range: 4-19) in Group 2. The total administered dose of G-CSF in Group 2 was significantly reduced by 40% ( P<0.0001). The delayed start of G-CSF after HDAC/MI in ALL achieves the same clinical benefit compared to the earlier initiation of G-CSF. The reduction of treatment costs by reducing the total G-CSF dose may be important in future treatment with this hematopoietic growth factor.