Study of the pathophysiology of infection requires the use of animal models in order to separate the many variables. Meaningful controlled studies of septic patients are difficult because of the diversity of diseases, different organisms, and multiplicity of variables encountered in the clinical setting. Endotoxin animal models, although reproducible, do not adequately simulate many of the circulatory and metabolic alterations produced by sepsis. Numerous sepsis models utilizing bacterial infection have been developed without any standardization of approach and there are often problems of reproducibility. When we began to study sepsis, we found that many of the proposed and reported small animal models of sepsis could not be reproduced in our laboratory. This led us through a long period of trial and error in order to develop a reproducible and satisfactory small animal sepsis model. Because of this, and the need for review and standardization, previously used models of sepsis such as the administration of endotoxin, intravenous infusion of live organisms, the administration of fecal material into the peritoneal cavity, the placement of infected foreign material into the soft tissues of the extremity, and surgical operations that partially destroy the normal barriers of the gastrointestinal tract are reviewed. Modifications of an existing sepsis model in the rat are also presented, and criteria for future models for the study of sepsis such as reproducibility of the model, clinical signs of sepsis, positive blood cultures, and alterations of vital signs are proposed.