Abstract: Near infrared spectroscopy (NIR) was used to detect citric acid and L-malic acid in pear juice quantitatively.The prediction models of citric acid and L-malic acid were optimized by combing 6 pretreatment methods with 3 mathematical correction components.The NIR spectra of 70 pear juice samples were collected in spectral range of 1 100~1 650 nm using an near infrared spectrometer, and the true content of citric acid and L-malic acid was measured by high performance liquid chromatography (HPLC).First, the original spectra were optimized with multiplication scatter correction (MSC), first derivative, second derivative and so on.Then, the prediction models of citric acid and L-malic acid content in pear juice were established using principal component regression (PCR), modified partial least square (MPLS), and partial least square (PLS), respectively.Finally, the prediction models were optimized and validated by eliminating outlier samples. The results indicated that the established mathematical model by MSC coupled with PLS, SNV and detrend coupled with MPLS had good predictive ability for citric acid and L-malic acid.The correlation coefficient of the best prediction model after optimized in prediction sets was 0.985, the minimum standard error of prediction (SEP) was 0.039%, and the maximum standard deviation ratio (SDR) was 3.46.The results showed that NIR spectroscopy could be used as a reliable, rapid, nondestructive and simple method to determine the content of citric acid and L-malic acid in pear juice.The effective and feasible method would help detect adulteration pear juices.