Methods for selecting a candidate drug for treating sepsis are disclosed. The methods involve labeling a sepsis-causing pathogen with a reporter and monitoring the progress of infection by detecting levels of the reporter in animals treated with test compounds or drugs. The comparisons may be made between experimental and control animals, as well as within a single animal or group of animals. Also disclosed is a method for predicting an expected time of death of an experimental animal in a model system of sepsis using data generated in the initial part of the experiment.