Updated project metadata. Purpose There are great demands for identifying markers of major depressive disorder (MDD), which is a common mental illness with a prevalence of approximately 6%. Finding potential markers to aid MDD diagnosis is in high demand. Experimental design In this study, combination of the salt-out assisted liquid-liquid extraction (SALLE) pretreatment method and a nontargeted peptidomics approach based on nano-LC-Orbitrap/MS is primarily employed to discover the candidate peptide biomarkers from the plasma of 238 subjects. Results A large number of peptides are enriched and identified from the plasma samples, of which 42 peptides show significant differences between MDD patients and controls by univariate statistical analysis. A diagnostic model combined four peptide markers (P1, P9, P17, P29) is established by binary logistic regression analysis, yielding an overall prediction accuracy of 91.7% and 82.2% in the discovery and validation sets, respectively. Conclusions and clinical relevance In conclusion, the good performance of diagnostic model in both discovery and validation sets demonstrates the robustness of peptide markers panel. It is very valuable for quantification the absolute content of four peptides and further verification.