The precondition for the antineoplastic effect of ifosfamide (ifo) is the oxidation of the oxazaphosphorine ring system, which contains a chiral centre at the phosphorous atom. This "ring oxidation" leads to the formation of alkylating mustard via several steps. A second metabolic pathway produces the cytostatically inactive metabolites 2- and 3-dechloroethyl-ifosfamide (2-d- and 3-d-ifo). The urinary excretion of the optical isomers of unmetabolised ifo and of 2- and 3-d-ifo, which represents the amount of ifo that has not been activated, was investigated by capillary gas chromatography for 18 treatment cycles in 14 children on various therapeutic schedules. The total cumulative excretion in 12 completely sampled cycles ranged from 27% to 50% of the ifo dose. Between 14% and 34% of the dose could be detected as ifo; 9% to 29%, as 3-d-ifo; and 2% to 8%, as 2-d-ifo. At 24 h after the end of therapy, excretion was nearly complete. Without exception, slightly more R-ifo (53%-61%) than S-ifo was excreted. S-2-d-ifo (50%-73%) was the main 2-d-metabolite. S-3-d-ifo (deriving from R-ifo) predominated in 6 of 14 children and R-3-d-ifo, in 8. Enantiomer-specific excretion increased after the end of infusion (up to 73% for R-ifo and 27% for S-ifo). We demonstrated stereospecific metabolism of ifo in children, with two different patterns of side-chain oxidation being observed. There was no evidence of important stereospecific ring oxidation in most children. A benefit should not be expected from the therapeutic application of pure enantiomers.