OBJECTIVE The aim of this study was to develop a limited sampling strategy (LSS) to estimate the area under the concentration-time curve (AUC) of gefitinib using data from 18 patients with non-small-cell lung cancer. METHODS On day 14 after beginning daily therapy with 250 mg of gefitinib, plasma samples were collected just before (C(0h), 24 hours after the 13th administration) and 1, 2, 4, 6, 8, 12, and 24 hours (C(nh)) after gefitinib administration and were analyzed by high-performance liquid chromatography. RESULTS The predicted AUC from 0 to 24 hours (AUC₀₋₂₄) from the single time point of C(12h) showed the highest correlation with the measured AUC₀₋₂₄ of gefitinib (AUC₀₋₂₄ = 20.0 · C(12h) + 1348.0; r² = 0.9623; P , 0.0001). The 95% confidence intervals of the slopes and intercepts of the formulae obtained by bootstrap analysis indicated acceptable accuracy and robustness in the prediction of AUC₀₋₂₄ using C(0h), C(1h), C(12h), and C(1h) + C(12h). The median AUC₀₋₂₄ and C(0h) of gefitinib in patients with diarrhea (n = 8) were higher than those without diarrhea (n = 10) (15,043 versus 8918 ng·h·mL⁻¹, respectively, P = 0.0164 and 542 versus 261 ng/mL, respectively, P = 0.0263). The area under the receiver operator curve was 0.813, giving the best sensitivity (75%) and specificity (90%) at a C(0h) threshold of 398 ng/mL. CONCLUSIONS Use of the C(12h) single time point is recommended for the gefitinib AUC₀₋₂₄ predictive equation; however, total estimation of the AUC₀₋₂₄ of gefitinib at the single point of C(0h) was also precise. C(0h) monitoring of gefitinib might be beneficial during gefitinib therapy, because of a high variability in gefitinib exposure among patients taking 250 mg. Further examination of the correlation between clinical evaluation and the gefitinib exposure is necessary.