Abstract: OBJECTIVE To construct a rapid prediction method for the determination of total polysaccharide content in Pteridium aquilinum by near-infrared spectroscopy. METHODS A total of 140 samples were collected, and the infrared spectrum in the range of 11500-4000 cm^-1 was collected by near-infrared spectroscopy technology through near-infrared diffuse reflection. The total polysaccharide content was determined by anthrone-sulfuric acid method as the actual value. The quantitative analysis model was established using the partial least squares method. The effects of different spectral preprocessing methods (including constant offset elimination, straight line subtraction, max min normalization, multiplication signal correction, standard normal variable, first derivative, second derivative, multiple scattering correction+first derivative method and standard normal variate + first derivative method) on the quantitative analysis model were compared to determine the best spectral pretreatment method, and to optimize the band of modeling. RESULTS The results show that when the standard normal variable method was used as the spectral data preprocessing method, 4450-4300 cm^-1, 5250-4700 cm^-1, 7500-5750 cm^-1, 10000-9000 cm^-1 were used as the interval bands, this model had the best performance, detection model R² and RMSECV were 0.985 1 and 0.129%, respectively. Through complete external validation, the average predicted recovery rate of the model was 100.13%. CONCLUSION The established rapid prediction model can accurately determine the total polysaccharide content of pteridium aquilinum, and can provide a candidate method for rapid determination of the main components of other traditional Chinese medicine.