To establish a rapid quantitative analysis method for the quality control of Cortex Phellodendri extraction by using near-infrared (NIR) spectroscopy and improve the prediction performance and robustness of the model by using Bagging partial least square regression (Bagging-PLS).Methods:NIR spectroscopy was collected online, and a partial least square (PLS) model was developed for determination the content of berberine hydrochlorid with HPLC method as referential method. Besides, bagging algorithm was adopted for improving the prediction performance and robustness of the model.Results:The RMSEC and RMSEP were 0.7429 and 0.9614, the correlation coefficient of the calibration model was 0.9866, and the correlation coefficient of the prediction was 0.9770. The mean PMSEP of the Bagging-PLS model was 0.4136 and the correlation coefficient of the prediction was 0.9983.Conclusion:In this study, NIR spectroscopy was successfully applied to Cortex Phellodendri extraction process, which can rapidly determine the content of berberine hydrochloride. In addition, the Baggting-PLS provided a better performance and prediction accuracy than single PLS model. This indicates that Baggting-PLS could potentially be used for a better predictive control of the extraction process.