Updated project metadata. Background: Recent advances in proteome analysis technology and mass spectrometry have made it possible to identify and quantify many proteins at once. The aim of this study was to identify proteins associated with severe burn pathology using mass spectrometry and to elucidate a novel molecular pathology classification that will be clinically useful. Methods: In this single-centre, retrospective, observational study, blood samples were collected from patients with severe burns at two time points, on the day of injury and 1 week after injury. Proteins were measured using mass spectrometry, and then prognosis-related proteins were extracted via comparison of 28-day survivor and non-survivor groups. Enrichment and ROC (receiver operating characteristics) analyses were performed to evaluate the extracted proteins. Subsequently, these proteins were used to perform latent class analysis. Findings: Measurements were performed on 83 burn patients. In the non-survivor group, ten proteins were significantly altered on the day of injury, and these were associated with multiple metabolic and response processes to toxins. Of these ten proteins that underwent ROC curve analysis with 28-day death as the objective variable, HBA1, TTR, and SERPINF2 on the day of injury had an area under the curve >0.8. Latent class analysis of these three proteins classified them into three molecular pathotypes. Interpretation: Comprehensive mass spectrometry of plasma revealed that ten proteins were associated with prognosis in severe burns. Molecular pathotypes based on HBA1, TTR, and SERPINF2 were significantly associated with patient outcomes.