Abstract: A method for rapid classification of toothpaste by differential Raman spectroscopy combined with K-means clustering method. 37 toothpaste samples numbered were smeared on slides and dried for scanning by differential Raman spectrometer. Functions of na.omit and scale in databases of fpc, factoextra and cluster in R language software were used to standardize the differential Raman spectral data of 37 toothpaste samples, and the elbow method and Gap Statistic algorithm were used to optimize the cluster number. Under the optimal cluster number of 4, the 37 toothpaste samples were classified by K-means clustering method and verified by hierarchical cluster analysis. It was showed that 37 toothpaste samples were divided into 4 categories, and the classification results of the two methods were consistent.