Preliminary experiments established that a 0.5-ml inoculum that is introduced directly into the stomach of mice was cleared rapidly into the small intestine. Bicarbonate buffer, but not skim milk, protected such an inoculum from stomach acid until at least 90% of it had entered the small intestine. Passage and survival of various Escherichia coli strains through the mouse gut were tested by introducing a buffered bacterial inoculum directly into the stomach, together with the following two intestinal tracers: Cr(51)Cl(3) and spores of a thermophilic Bacillus sp. Quantitative recovery of excreted bacteria was accomplished by collecting the feces overnight in a refrigerated cage pan. The data show that wild-type E. coli strains and E. coli K-12 are excreted rapidly (98 to 100% within 18 h) in the feces without overall multiplication or death. E. coli varkappa1776 and DP50supF, i.e., strains certified for recombinant DNA experiments underwent rapid death in vivo, such that their excretion in the feces was reduced to approximately 1.1 and 4.7% of the inoculum, respectively. The acidity of the stomach had little bactericidal effect on the E. coli K-12 strain tested, but significantly reduced the survival of more acidsensitive bacteria (Vibrio cholerae) under these conditions. Long-term implantation of E. coli strains into continuous-flow cultures of mouse cecal flora or into conventional mice was difficult to accomplish. In contrast, when the E. coli strain was first inoculated into sterile continuous-flow cultures or into germfree mice, which were subsequently associated with conventional mouse cecal flora, the E. coli strains persisted in a large proportion of the animals at levels resembling E. coli populations in conventional mice. Metabolic adaptation contributed only partially to the success of an E. coli inoculum that was introduced first. A mathematical model is described which explains this phenomenon on the basis of competition for adhesion sites in which an advantage accrues to the bacterium which occupies those sites first. The mathematical model predicts that two or more bacterial strains that compete in the gut for the same limiting nutrient can coexist, if the metabolically less efficient strains have specific adhesion sites available. The specific rate constant of E. coli growth in monoassociated gnotobiotic mice was 2.0 h(-1), whereas the excretion rate in conventional animals was -0.23 h(-1). Consequently, limitation of growth must be regarded as the primary mechanism controlling bacterial populations in the large intestine. The beginnings of a general hypothesis of the ecology of the large intestine are proposed, in which the effects of the competitive metabolic interactions described earlier are modified by the effects of bacterial association with the intestinal wall.