A prolonged cytotoxic effect of 5-azacytidine (aza-CR) on leukemic colony-forming units (LCFU) was observed in mice with transplanted L1210 leukemia. LCFU showed rapid reaccumulation in the marrow 12 hr after injection of 0.1 mg of aza-CR per mouse. However, after 0.5 mg of aza-CR, repopulation was delayed for at least 6 days. Experiments were performed to determine the mechanism of this prolonged antileukemic effect. Suspensions of leukemic marrow prepared from mice treated 4 days previously with 0.5 mg of aza-CR were exposed to [3H]thymidine in vitro in order to kill cells in S phase. Suspensions exhibited a 40% reduction in LCFU, indicating the prolonged effect was not due to cell cycle progression delay. Mice given whole-body irradiation prior to receiving L1210 demonstrated the same delayed repopulation following the high dose of aza-CR as nonirradiated mice, suggesting that the effect was likely not due to an immune reaction. aza-CR, when given to normal mice as long as 2 days prior to leukemic transplantation, was able to prolong the survival of leukemic mice, but not when given at longer intervals. Administration of aza-CR to mice 1 day or 1 hr prior to leukemic transplantation resulted in decreased LCFU survival as well as delayed repopulation of LCFU; the rate of repopulation was not changed. This indicated a prolonged residual activity of the drug, but not sufficient to explain the total in vivo suppression. In contrast, administration of aza-CR to leukemic mice suppressed repopulation of a subsequent leukemic transplant for 4 days, even when the cells were given 2 days after the aza-CR. Cytidine was partially able to reverse the delayed repopulation of LCFU when given 1 day after aza-CR, but it was unable to reverse the phenomenon 2 days after aza-CR. Therefore, a high dose of aza-CR produces a prolonged antileukemic effect which is probably mediated by continued availability of an aza-CR metabolite. Since this effect is more pronounced in leukemic mice than in nonleukemic mice, the pharmacokinetics of high doses of aza-CR probably differ in normal and leukemic mice.