In order to explore the Raman fingerprint of food-borne pathogenic bacteria spores for rapid identification. In this study, the spores of C. perfringens, C. difficile and B. cereus were used as the research objects. The SERS technology of AgNPs synthesized by the sodium citrate reduction method was used to detect the Raman spectroscopy of food-borne pathogenic bacteria spores and analyze the similarities or differences among different spores. The SERS spectra of three kinds of food-borne pathogenic bacteria spores were combined with principal component analysis (PCA) and hierarchical cluster analysis (HCA) for comparative analysis to identify different species of food-borne pathogenic bacteria spores. The results showed that the SERS spectra of different food-borne pathogen spores had sufficient specificity and reproducibility. In the SERS spectra of spores, the number and intensity of Raman vibration peaks of Ca²⁺-DPA were dominant, and the Raman vibration peaks were located at 657~663, 818~820, 1 017, 1 389~1 393, 1 441~1 449 and 1 572~1 576 cm^-1. The intensity of six characteristic peaks of Ca²⁺-DPA in SERS spectra of C. difficile spores were higher than that of C. perfringens spores and B. cereus spores, followed by C. perfringens spores. The Raman peak intensity of Ca²⁺-DPA at 1 017 cm^-1 (Ca²⁺-DPA) of the three spores was the highest, and the difference was noticeable, which was the main characteristic peak of Ca²⁺-DPA and the main characteristic peak of the three spores. In addition, C. perfringens spores showed unique Raman peaks at 936 cm^-1 (N—C stretching of phospholipid), 1 294 cm^-1 (CH₂ deformation vibration of lipid), 1 609 cm^-1 (tyrosine of protein) and 1 649 cm^-1 (amide Ⅰ of protein). C. difficile spores showed unique Raman peaks at 890 cm^-1 (=C—O—C=stretching). PCA analysis showed that the variance contribution rates of PC1 and PC2 were 51.10% and 39.70%, respectively, and the cumulative contribution rate was 90.8%, which could effectively distinguish all samples. HCA analysis indicated that the SERS spectra of the three spores were divided into three clusters, and each cluster of the three spores had no cross-interference. The combination of multivariate statistical analysis effectively realized the distinction among the three spores and distinguished the distinction between Clostridium spores and Bacillus spores, providing an effective means for food safety control.