Fourteen poor risk acute myeloid leukemia (AML) patients were treated with G-CSF prior (from day 0) and during chemotherapy with fludarabine and Ara-C from day 1 to day 5 (the FLAG regimen). Several biological parameters were monitored on the blast population: multidrug resistance (MDR) functional expression by rhodamine-123 efflux (Rhd-E), cell cycle changes, induction of apoptosis and leukemic clonogenic cell growth (CFU-L). The mean basal Rhd-E value was 14.4% (range 0-51.2), and 12/14 patients exhibited a dye efflux > 4%, efficiently blocked by the MDR-reversal agent cyclosporin A. After 24 h of G-CSF administration, cell cycle studies showed in bone marrow (BM) samples a significant mean increase in S phase (p = 0.04) and in RNA content of G1 cells (p = 0.01), coupled to a significant increase in apoptosis (p = 0.02). Clonogenic cell growth analysis showed a twofold increase in BM CFU-L in 6 of the 14 cases tested. When G-CSF activity was assessed without the addition of exogenous growth factors (autonomous proliferation), a significant increase (p = 0.02) in CFU-L was found only in patients who achieved a complete remission (CR); these patients were also characterized by lower S-phase values at diagnosis. Eight of the 14 patients treated achieved CR, but the median response duration was three months, and only two cases are still in CR. The FLAG regimen can thus induce remission in poor risk AML patients. The responses, however, are short, suggesting that resistant cells are not efficiently affected by either the use of agents not involved in the MDR-efflux mechanism or by the G-CSF priming strategy. Other post-induction therapies need to be considered in further approaches.