Thirty-five samples of bone marrow (BM) from 17 patients (pts) with ALL and 18 pts with AML (aged 9 m-20 yrs, median 7.7 yrs) were obtained. Using MTT-assay the sensitivity of LB to Ara-C, VP-16, DOX, G(GM)-CSF and their combinations was measured. LC50 was higher in pts with AML than with ALL: to Ara-C 1.94-fold (p < 0.05), to VP-16 1.62-fold (p = 0.2), to DOX 3.9-fold (p < 0.05). Incubation with G-CSF increased the viability of ALL and AML LB--104.3% and 104.1% respectively (the viability of leukemic cells without CSF accepted as 100%). Incubation with GM-CSF decreased the viability of ALL LB (96.5%) and increased the viability of AML LB (139.1%) (p = 0.08). Combining Ara-C with G- or GM-CSF resulted in equal or increased LC50 (compared with LC50 of Ara-C alone) in 100% cases of AML. For ALL: LC50 of "Ara-C+G-CSF" was equal or increased in 63.6% cases; LC50 of "Ara-C+GM-CSF"-in 62.5%. For VP-16 and DOX all pts (ALL, AML) except two had equal or increased LC50 of "CH+CSF" (compared with LC50 of CH alone). These data show: 1) AML LB were less sensitive to the investigated CH than ALL LB. 2) The LC50 of "CH+CSF" was equal or increased compared to the LC50 of CH for the absolute majority of cases with VP-16 and DOX. The same results were obtained with AML and in about 60% cases of ALL. The effect of the increasing of cytototoxity of CH in presence of CSF probably exists mostly at higher concentrations of CH than those that can be achieved in clinical practice.