The limited sampling model (LSM) offers a means of estimating the area under the concentration-time curve (AUC) from only two timed plasma concentrations. In this study, pharmacokinetic profiles were simulated for 23 patients treated with amonafide, using each patient's individual pharmacokinetic parameters. Data were simulated for a dose of 250 mg/m2 administered over 1 h. The initial 15 patients formed the training data set. Based on the training data set, five different LSMs were generated, with the multiple r ranging from 0.92 to 0.98. A single model was selected as optimal: AUC (micrograms min/ml) = 292.9 (min) C45 (micrograms/ml) + 3262 (min) C1440 (micrograms/ml) + 21.8 (micrograms min/ml) dose (mg/m2)/250 mg/m2 where C45 = 45-min plasma concentration and C1440 = 24-h plasma concentration. This model was revalidated on a second test data set of seven patients actually treated with a 1-h infusion. The relative root mean square predictive error was 15.8%, acceptable for most clinical uses. We conclude that the LSM is a powerful tool for estimation of the AUC in a large patient population. The LSM may facilitate population pharmacodynamic studies in conjunction with Phase II trials.