We have studied the pharmacokinetics of mitoxantrone in cancer patients. Two regimens were used: eight women (10 kinetics) received a 10 min i.v. infusion of 12 mg m-2 of mitoxantrone; seven women (seven kinetics) received high-dose mitoxantrone associated to high-dose alkylating agents and underwent autologous bone marrow transplantation (BMT). High-dose mitoxantrone was administered according to two different protocols. The drug was quantified in plasma with an HPLC assay and pharmacokinetic analysis was performed with the APIS software. Mitoxantrone pharmacokinetics were best described by an open two- (six kinetics) or an open three compartment model (11 kinetics). A large interindivual variability was observed in pharmacokinetic parameters. In the first group of patients, mean +/- s.d. values of clearance, half-life and total distribution volume were 21.41 +/- 14.59 1 h-1, 19.83 +/- 23.95 h, 165.89 +/- 134.75 1 respectively. In the high-dose group, these values were 21.68 +/- 7.30 1 h-1, 50.26 +/- 20.62 h, 413.70 +/- 194.81 1 respectively. Results showed that identification through the open 2-compartment model is certainly related to the small number of late time-points. We therefore think that mitoxantrone pharmacokinetics is generally best described by an open 3-compartment model. Clearance values showed that there was no saturation in mitoxantrone elimination, even at the highest doses. Terminal elimination half-life was probably underestimated because of the lack of late time-points in some kinetics. The half-life is long for patients receiving high-dose mitoxantrone (mean value was 50 h) and it would be hazardous to perform BMT too early after mitoxantrone infusion. Mitoxantrone metabolites were detected in the plasma of five patients receiving high-dose mitoxantrone and in one with hepatic impairment.