Background Optimization of the area under the concentration–time curve (AUC) of busulfan, an essential component of conditioning regimens, improves the outcomes in patients undergoing hematopoietic stem cell transplantation (HSCT). Traditional sampling methods for calculating AUC require multiple sampling. Objective To establish a limited sampling strategy for predicting the AUC0-12 of intravenous busulfan for Chinese adult patients prior to HSCT. Methods The pharmacokinetics of twice-daily intravenous busulfan was studied in forty-five Chinese adult patients. Limited sampling models were established by the multiple linear regression analysis. The prediction error (PE) and the absolute prediction error (APE) were calculated to evaluate predictive accuracy. The agreement between the predicted and actual AUC0-12 was investigated by the Bland–Altman analysis. The accuracy and robustness of the models was validated by the bootstrap analysis. Results The AUC0-12 values of the 1st and 7th doses of busulfan were 1491 ± 403.7 and 1908 ± 518.5 μmol L−1 min, respectively. The 2-sample model suggested that the predicted AUC0-12 of twice-daily intravenous busulfan could be calculated using the following equation: AUC0-12 = 40.017 + 0.955 × C3 + 1.088 × C6 with r2 = 0.919. The mean PE and APE of the model were 0.52 ± 7.67 and 6.32 ± 4.27%, respectively. Conclusion The 2-sample model is an effective and reliable approach to predict the AUC0-12 of twice-daily intravenous busulfan in Chinese adult patients.