Abstract— This paper investigates the use of the median filter (MF) for the quantitative analysis of the NIR spectra. MF is a nonlinear low pass filter used widely in digital image processing for smoothing and at the same time the edges were preserved. MF is more robust and less sensitive to outliers compared to the Moving average filters. The use of MF for the quantitative analysis of the NIR spectra has not been previously evaluated in the field of chemometrics. In this work, MF is used as a pre-processing method to the Partial Least Squares Regression (PLSR) model. The effect of using MF has been evaluated and compared to the Moving Window Average (MWA) filter and Savitzky-Golay filter by computing the Standard Error of Prediction (SEP), R-squared (R^2), and Mean Absolute Percentage Error (MAPE). The model is validated using different mixtures composed of glucose, urea and triacetin dissolved in a phosphate buffer solution. The results show that using MF combined with the MWA filter and PLSR improves SEP of the PLSR model from 35.6 mg/dL to 18 mg/dL.