Objective: To analyze the feasibility of using Near-infrared (NIR) spectroscopy for quality control of liquid-liquid extraction procedure of Cardeniae Fructus in the preparation of reduning injection. Method: 147 extraction stream samples from seven batches were collected and NIR off-line spectra were acquired by Luminar 5030 spectrometer. The contents of geniposide (GS) and soluble solid content (SSC) were determined by the reference method. The partial least square(PLS) algorithm was used to build GS and SSC NIR models and predict the contents in prediction set. Result: The R2 of the established GS and SSC models were 0.987 2 and 0.994 7 as well as the RMSEC was 1.460 9 and 2.367 7

which indicated the models had good performance. Besides

the predictive results of 20 samples were satisfactory. The R2 was 0.980 7 and 0.986 1

the RMSEP was 1.827 5 and 7.307 7 and the RSEP was 3.08% and 5.29% for GS and SSC

respectively. Both of the RSEP values were lower than 6%

which can satisfy the quality control standard in the traditional Chinese medicine production process. Conclusion: The established models could precisely predict the content of GS and SSC. The results demonstrated that the NIR spectroscopy might be a suitable and alternative technology for monitoring the liquid-liquid extraction procedure of Gardeniae Fructus.