Background: The mortality rate from sepsis has improved little over the past two decades. One reason for this has been the use of flawed or inappropriate experimental models in preclinical sepsis studies.
Methods: A literature review of animal models of sepsis was performed following a Medline search based on the following medical subject headings: disease models, endotoxin, inflammation, peritonitis and sepsis. Additional references were identified from the papers identified in the search.
Results and conclusion: Many animal models of sepsis have been described but none has proved to be superior. Extrapolation of results from endotoxicosis or bacterial infusion models should be regarded with caution. Peritonitis models should be accepted as the 'gold standard' but the use of appropriate virulent bacterial species needs to be ensured. A standardized panel of animal models for the preclinical assessment of immunomodulatory agents should be established, including at least one immuno- suppressed model to simulate the immunocompromised patient with sepsis. A uniform and valid definition of sepsis applicable to both small and large animal species is required.