Enteral diets with different protein content were tested to determine their effect on outcome in a model of protracted bacterial peritonitis. Hartley guinea pigs were provided with gastrostomies, and 1 week later, osmotic pumps were implanted into the peritoneal cavity to allow for continuous release of live bacteria over the course of 1 week. Three days after pump implantation, the animals began receiving isocaloric enteral diets that contained 5%, 10%, 15%, or 20% of total calories as protein. After 2 weeks of observation, the survivors were killed. All animals lost weight during the 2-weeks period, but there was no difference in weight lost. Nitrogen balance correlated with dietary protein. The mortality rate was significantly higher in the groups that received 15% and 20% of total calories compared with the group that received 5% (p less than 0.05). Although dietary protein in the 5% group was insufficient for meeting the nutritional needs of the animal, survival was best in this group. Possible explanations are that protein restriction in this model may either augment host defence or impair bacterial virulence.