This study was aimed to find new biomarkers for diagnosis and prediction of prognosis of sepsis. Serum samples from nonsurvivor, survivor, and control groups were obtained at 12 h after the induction of sepsis and labeled with isobaric tags (iTRAQ) and then analyzed by two-dimensional liquid chromatography and tandem mass spectrometry. Protein identification and quantification were obtained using mass spectrometry and the ProteinPilot software. Bioinformatics annotation was performed by searching against the PANTHER database. Enzyme-linked immunosorbent assays were used to further confirm the protein identification and differential expression. A logistic regression was then used to screen the index set for diagnosis and prognosis of sepsis. We found that 47 proteins were preferentially elevated in septic rats (both nonsurvivors and survivors) compared with the control rats, and 28 proteins were preferentially elevated in the NS rats as compared with the S group. Several biomarkers, such as multimerin 1, ficolin 1, carboxypeptidase N (CPN2), serine protease 1, and platelet factor 4, were tightly correlated with the diagnosis of sepsis. Logistic regression analyses established multimerin 1, pro-platelet basic protein, fibrinogen-α, and fibrinogen-β for prognosis of sepsis.