The development and validation of a high-performance liquid chromatographic (HPLC) assay for determination of busulfan concentrations in human plasma for pharmacokinetic studies is described. Plasma samples containing busulfan and 1,6-bis(methanesulfonyloxy)hexane, and internal standard, were prepared by derivatization with sodium diethyldithiocarbamate (DDTC) followed by addition of methanol and extraction with ethyl acetate. The extract was dried under nitrogen and the samples reconstituted with 100 microl of methanol prior to HPLC determination. Chromatography was accomplished using a Waters NovaPak octadecylsilyl (ODS) (150 x 3.9 mm I.D.) analytical column, NovaPak ODS guard column, and mobile phase of methanol-water (80:20, v/v) at a flow-rate of 0.8 ml/min with UV detection at 251 nm. The limit of detection was 0.0200 microg/ml (signal-to-noise ratio of 6) with a limit of quantitation (LOQ) of 0.0600 microg/ml for busulfan in plasma. Calibration curves were linear from 0.0600 to 3.00 microg/ml in plasma (500 microl) using a 1/y weighting scheme. Precision of the assay, as represented by C.V. of the observed peak area ratio values, ranged from 4.41 to 13.5% (13.5% at LOQ). No day-to-day variability was observed in predicted concentration values and the bias was low for all concentrations evaluated (bias: 0 to 4.76%; LOQ: 2.91%). The mean derivatization and extraction yield observed for busulfan in plasma at 0.200, 1.20 and 2.00 microg/ml was 98.5% (range 93.4 to 107%). Plasma samples containing potential busulfan metabolites and co-administered drugs, which may be present in clinical samples, provided no response indicating this assay procedure is selective for busulfan. This method was used to analyze plasma concentrations following administration of a 1 mg/kg oral busulfan dose.