The clinical pharmacokinetics of the antineoplastic agent irinotecan (CPT-11) are associated with substantial interpatient variability. The degree to which this variability in CPT-11 exposure impacts upon the response and toxicity of the drug has not yet been properly determined. In general, the area under the plasma concentration-time curve (AUC) is an appropriate indicator of exposure, but requires collection of up to 17 timed blood samples. This presents difficulties if large-scale population samplings are required. The present study involved the development and validation of a strategy to estimate the AUCs of the lactone and total (i.e. lactone plus carboxylate) forms of CPT-11 and its active metabolite SN-38 from a limited number of blood samples in patients co-treated with cisplatin. Using data from 24 patients, univariate and multivariate regression analyses were employed to generate the models. The best predictive models for simultaneous estimation of CPT-11 and SN-38 AUCs were obtained with three time points at 0.5, 1.67 and 5.50 h after start of the 90 min i.v. infusion of CPT-11. The models were tested separately in another group of 24 patients receiving the same combination treatment. This validation set demonstrated that CPT-11 and SN-38 AUCs after standard dose administration could be predicted sufficiently unbiased and precisely with three timed samples to warrant clinical application.