All limited sampling models so far proposed to determine the area under the concentration-time curve (AUC) of anticancer drugs can be applied only to the dosing/sampling schedule used to obtain the model. The authors have developed a new method to predict the AUC of 5-fluorouracil (5-FU) after rapid intravenous bolus administration of various doses, using as few as two plasma drug concentrations. The 5-FU AUC (AUC(true)) was first determined in 20 patients receiving adjuvant therapy for colorectal cancer, based on nine plasma drug concentrations obtained at 0, 2.5, 5, 10, 15, 20, 30, 45, and 60 minutes after drug administration. Ten patients received 375 mg/m(2) 5F-U + 100 mg/m(2) leucovorin and 10 received 425 mg/m(2) 5-FU + 20 mg/m(2) leucovorin. The kinetics of 5-FU was described by either a one- or two-compartment linear model, as needed. The AUC was then recalculated (AUC(approx)) using a reduced number of plasma concentrations and a simple one-compartment model. The time combinations tested were 2.5, 5, 10, and 20; 2.5, 10, and 20; 5, 10, and 20; 5 and 20; and 2.5 and 20 minutes. The accuracy and precision of the method in predicting the AUC(true) were measured by calculating the mean prediction error (MPE%) and the mean absolute error (MAE%) of the AUC(approx). MPE% ranged between -0.8% and -8.3% and MAE% between 6.1% and 9.5%, depending on the time combination used. In general, all limited sampling models tested tended to underestimate the AUC(true) slightly, particularly when 5-FU kinetics followed a two-compartment model, but bias was still within acceptable limits. The best results were obtained with plasma concentrations measured at 2.5 and 20 minutes after drug bolus (MPE%, -0.8%; MAE%, 6.1%). Although 5-FU kinetics was dose-dependent, MPE% and MAE% were not significantly different between the two groups. These data show that 5-FU AUC can be reliably predicted by using just two plasma concentrations and a one-compartment model, even when different doses are used and plasma concentration decay is biexponential.