We hypothesized that the steady-state concentration of intracellular cytarabine 5'-triphosphate (ara-CTPss) in leukemia cells is proportional to the dose rate of cytarabine (ara-C) during continuous infusion. To evaluate this possibility, patients with acute myelogenous leukemia in relapse were treated with two sequential schedules of serially increasing ara-C dose rates over a total of 36 hours. Schedule I consisted of serial infusions of 250, 500, and 750 mg/m2 each over 12 hours. Subsequently, patients entered on schedule II received 500, 1,000, and 1,500 mg/m2 serially, each over 12 hours. Steady-state levels of ara-CTP were achieved within four hours after beginning ara-C infusion and, in separate studies of a single ara-C dose rate, were shown to be maintained beyond 36 hours. Four patients treated with schedule I and two patients treated with schedule II showed a linear dose rate-dependent increase-of ara-CTPss at all three dose rates. The cells of one patient on schedule I and two patients on schedule II had a dose rate-dependent ara-CTPss increase only over the first two dose levels, while no increase or lower ara-CTPss was observed at the third dose rate. The ara-CTPss of one patient on schedule II did not change. These results suggest that there is a proportionality between the continuous infusion dose rate of ara-C and the ara-CTPss in circulating leukemia cells within the dose range of 250 to 1,000 mg/m2 over 12 hours. This opens the possibility that pharmacologic determinations may be used to redirect the ara-C dose rate to achieve a desired ara-CTPss level in leukemia blasts during therapy.