Pantana phyllostachysae Chao (PPC) is one of the most important leaf-eating pests of bamboo forests in China. It has become a major factor threatening the health of Moso bamboo forest and restricting the high quality and sustainable development of bamboo industry. It also has the characteristics of group-occurring, periodicity, and extremely serious harm, etc. How to quickly and accurately detect the damage of the Moso bamboo forest is a problem that needs to be solved at this stage. Whereas remote-sensing products can support the quickly, accurate, and comprehensive monitoring of forest health. Therefore, Sentinel-2A MultiSpectral Instrument (MSI) data, with three bands at the red-edge position, was of great significance for pest and disease detection in forests. By screening 22 spectrally derived indicators (e.g. leaf abscission, greenness and water content) using ANOVA combined with recursive RFE, totally 10 features were finally obtained to identify PPC damage. Based on the above results, the XGBoost detection model was established to detect PCC damage with high recognition accuracy. The results showed that Sentinel-2A MSI bands 6, 7, 8, and 8a exhibited strong responses to PPC damage;the index constructed by the red-edge and near-infrared bands effectively reflected the damage to bamboo forests;the overall detection accuracy of model was 83.70% compared with 94.72%, 72.06%, 79.77%, and 92.41% for ‘healthy’, ‘mildly damaged’, ‘moderately damaged’, and ‘severely damaged’ categories, respectively. These results indicated that the XGBoost detection model provided valuable support for the large-scale monitoring of pest damage to Moso bamboo forests.