In phase I and II clinical trials, the deoxycytidine analogue 2',2' difluorodeoxycytidine (dFdC, gemcitabine) has shown promising antitumor activity in leukemia as well as in solid tumors. Preclinical and clinical studies of gemcitabine suggested that myelosuppression was the dose-limiting toxicity. The present investigations were designed to test the effect of continuously administered gemcitabine on the in vitro clonal growth of normal CD34(+) cells isolated from peripheral blood and the promyelocytic cell line, HL-60. For this purpose, CD34(+) and HL-60 cells were cultured in methylcellulose in the continuous presence of 0.1-16 nM of gemcitabine. The results show a dose-dependent inhibition of colony growth of normal as well as leukemic cells. However, HL-60 cells were up to 12-fold more sensitive towards gemcitabine than normal progenitors. For rescue experiments, the natural pyrimidine deoxycytidine (dCyd) was added to CD34(+) and HL-60 cells simultaneously or with delay. Coadministration of 1mM dCyd to separate cultures resulted in complete restoration of colony formation capacity of CD34(+) and HL-60 cells. Delayed addition of 1 mM dCyd after 48 and 72 hours recovered up to 90% and 40%, respectively, of stem cell proliferation, whereas HL-60 cells remained substantially inhibited (4.5% +/- 3.5% versus 0%). Delayed addition after 48 and 72 hours protected about 80% and 50%, respectively, of myelopoietic and erythropoetic colony formation, whereas colony formation obtained from HL-60 cells remained significantly inhibited (9.6% +/- 4.17% versus 0%). These in vitro data suggest that there is a marked difference in the susceptibility of leukemic and normal CD34(+) cells to gemcitabine and that delayed administration of dCyd may further reduce the bone marrow cytotoxicity of gemcitabine without impairing its antitumor effect.