Abstract: In order to realize the accurate and rapid classification of food packaging paper, the method mentioned by the title was proposed. Differential Raman spectroscopy was used to analyze 48 food packaging paper samples of different brands and sources, and the original differential Raman spectral data of the samples were obtained. According to the differences in the main fillers in the food packaging paper samples, the samples were simply divided into two categories, and the class I samples containing talc were further divided into 4 groups (manual grouping). Principal component analysis was used to reduce the dimension of the original spectral data into 26 variables, and the Fisher discriminant analysis method in SPSS 26.0 software was used to verify the results obtained by manual grouping.The overall correct discrimination rate of the model discrimination was 94.3%, and it was turned out that the results of manual grouping was quite accurate. In order to reduce the error of manual grouping, the dimensionality-reduced data were analyzed with hierarchial-cluster analysis. When the squared Euclidean distance was 1, the class I samples could be divided into 7 types, and the results were verified by the Pearson correlation coefficient model. The method was non-destructive on the simple, fast and stable, which was helpful for the inspection of relevant trace evidence.