Many species of chili peppers have overlapping morphological characters and delimitation by visual descriptors in many cases fails to differentiate one species from another. In Peru, there are 413 accessions of native chili pepper and 296 accessions of rocotos conserved in the Germplasm Collections of the National Institute of Agrarian Innovation (INIA), of which five accessions (three species from three locations) were selected for the present metabolomic study. The Discrimination of the three species of native chili peppers and identification of biomarkers was performed using untargeted metabolomic approach based on profiling by UHPLC-HRMS and multivariate data analysis. The samples of fresh chili peppers (whole fruit) from Chincha area were used to construct an OPLS-DA model. To validate the biomarkers (identified 15 biomarkers, mainly flavonoids), an external validation set of the OPLS-DA model was constructed using Chiclayo and Huaral collection datasets. Consequently, the OPLS-DA based on Chincha samples model has a high predictive capacity demonstrating that the biomarkers have a high probability of continuity in any culture space, being successful in discriminating the species by untargeted metabolomics.