Early biomarkers of radiation injury are critical for triage, treatment, and follow-up of large numbers of people exposed to ionising radiation after terrorist attacks or nuclear accident. Operational monoparametric protein or amino acid biomarkers (amylase, Flt3-Ligand, citrulline) can help for the diagnostic of radiation exposure or injury. However, these biomarkers are not sufficient for a fast and accurate triage, and if individuals are assessed more than 48 h after exposure. The comparative proteomic approach represents a promising powerful tool for the discovery of new radiation biomarkers. In association with multivariate statistics, proteomic enables to measure the level of hundreds or thousands of proteins at the same time and identifies sets of proteins that can discriminate different groups of individuals.