The work aims to develop a methodcombining infrared spectroscopy with chemometrics to achieve rapid inspection and classification of express carton samples on site. 53 samples of express cartons were inspected by infrared spectroscopy and classified according to the differences of their main fillers, and then grouped by clustering method. According to the grouping results, random forest, multi-layered perceptron and Fisher discriminant analysis were established to realize the prediction of new sample groups. 53 express carton samples were divided into 3 categories, and then subdivided into 9 groups, and Fisher discriminant method could achieve more accurate prediction. The inspection method is rapid, non-destructive, accurate, and realizes the rapid inspection of express carton samples based on chemometrics, which provides a basis for public security organs to examine such material evidence.