Cytarabine is intracellularly activated and correlations have been established between the pharmacokinetic behaviour of active metabolites and their antileukemic effect. Recently, a good response to high-dose treatment of leukemias has additionally been attributed to a so-called low deamination phenotype of cytarabine inactivation. Consequently, these findings would support plasma level monitoring of cytarabine and its metabolite uracil arabinoside in high-dose cytarabine regimens. This pharmacokinetic study presents data attempting to reevaluate these observations. Thirty-seven patients were treated by 3-h high-dose cytarabine infusions (9 patients 1000 mg/m2, 28 patients 3000 mg/m2) as part of their treatment for acute leukemia. Serial blood samples during and post infusion were analysed for cytarabine (araC) and its deamination product uracil arabinoside (araU) using HPLC with UV-detection. Considerable interindividual variation was observed in end-infusion plasma concentrations of araC (1000 mg/m2: 2.1-fold, 3000 mg/m2: 5.5-fold) and araU (1000 mg/m2: 2.7-fold, 3000 mg/m2: 2.9-fold). The median ratio of end infusion concentrations araU/araC (on a molar basis) was 5.6 (S.D. 3.0), extreme ratio values were 2 and 14. No differences of the araU/araC ratio were found between the two dosages used. Minimum plasma araC concentrations at the end of infusion were 10.5 micromol/l and 22.0 micromol/l at a dose of 1000 and 3000 mg/m2, respectively. In our European study population a "fast" deamination phenotype of cytarabine (araU/araC ratio > 14) was not be observed.