Infected, neutropenic animals are used as experimental models to evaluate the relative efficacies of antimicrobial agents and host-pathogen-antibiotic interactions. In the past, these models used death as the study end point. Because of the concern about use of death as an end point, we evaluated the accuracy with which various signs of infection predicted mortality in a neutropenic guinea pig model of treated and untreated Pseudomonas aeruginosa sepsis. The potential surrogate markers studied included ruffled fur, respiratory distress, diarrhea, hunched posture, lethargy, abnormal neurologic movements (twitching, paralysis of a limb), inappetence for > 48 h, the inability to ambulate, and the inability of a supine animal to stand. In addition, we evaluated whether percentage of weight loss or change in daily food and water consumption were predictive of mortality. Animals were inspected for these signs at least every 4 h during the day and every 8 h in the evening. In treated and untreated animals, 100% of subjects that were unable to ambulate or to rise from the supine position died (positive predictive value for death was 100% for either sign). Guinea pigs that could not rise from a supine position expired between 1 and 8 h after this sign was observed. Those that could not ambulate died between 4 and 40 h after that sign was observed. In treated and untreated animals, none of the survivors manifested either sign of disease (100% specificity for each sign). However, 59% of untreated and 69% of treated animals that were ambulatory were found dead at the next observation period, underscoring the rapidity with which this infection progresses to death when it enters its final stage. No other signs of infection distinguished animals that survived or died. Thus, the inability of neutropenic, infected guinea pigs to rise from a supine position and the inability to ambulate were the only signs that accurately predicted death and, therefore, are the only signs that can be used as surrogates for death in this experimental model of P. aeruginosa sepsis.