A study was made of the effect of certain dietary regimens on the lactobacillus flora in the stools of mice and on their resistance to infection. Semi-synthetic diets with purified casein or wheat gluten as sole source of protein, gave rise to much smaller numbers of viable lactobacilli in the stools than did other diets containing unidentified natural products-as present for example in mixtures of whole wheat and whole milk, or in certain commercial pellets. Furthermore, one of the lactobacillus types with rhizoid morphology disappeared completely from the stools of animals fed the semi-synthetic diet. The change in the lactobacillus flora became apparent within a very few days after the animals had been shifted from the complex to the synthetic diet Moreover, this change was not completely reversible. Whereas the total numbers of lactobacilli increased when the animals were shifted back from the synthetic to the complex diets, the rhizoid lactobacilli which had disappeared completely from the stools reappeared only slowly or not at all. In twelve consecutive experiments the three diets which gave rise to the large numbers of lactobacilli in the stools also conferred on the mice a much higher resistance to experimental infection with Staphylococcus aureus and Klebsiella pneumoniae, than did the semi-synthetic diets. However, direct evidence has not yet been obtained that the two kinds of phenomena were causally related. Following administration of endotoxin there was a rapid and very large increase in the numbers of enterococci and coliform bacilli in mice fed the semi-synthetic casein diet, but not in those fed the pellets. In two preliminary experiments carried out with another colony of mice, not pathogen-free, it was also found that the rhizoid type of lactobacilli disappeared from animals fed the semi-synthetic casein diet while enterococci and coliform bacilli progressively increased in numbers under the same conditions. The dietary effects on the lactobacillus flora, and on resistance to experimental infection, were equally pronounced whether the mice were housed in individual cages on wire grids, or grouped in larger cages with wood shavings as litter. This was true even if the bedding was changed only once weekly and became therefore grossly soiled.