In the last decades several preclinical models for sepsis have been used to study the pathophysiologic processes during sepsis. Although these studies revealed promising immunomodulating agents for the treatment of sepsis, clinical trials evaluating the efficacy of these new agents in septic patients were disappointing. It should be realized that most of the preclinical models for sepsis lack a localized infectious source from which the infection disseminates. Studies on the effects of several immunomodulating strategies have demonstrated strikingly opposite results when using models for sepsis with a more natural route of infection, such as pneumonia, and when using models for sepsis lacking an infectious focus. In this review we will compare models for sepsis and models for pneumonia. We advise to use a combination of models, including models for sepsis and models for localized infections, to test new immunomodulating strategies before starting any clinical trial evaluating a new immunomodulating therapy.