By studying Bartonella prevalence in rodent communities from 23 geographic sites in the western United States and one site in northern Mexico, the present study focused on the effects of rodent community diversity (measured by richness and Shannon index) and composition on prevalence of Bartonella infections. The analysis showed negative correlations of Bartonella prevalence with rodent richness and Shannon index. Further, Bartonella prevalence varied among rodent genera/species. Three models were applied to explain the observations. (1) Within-species/genus transmission: Bartonella strains usually are host-specific and adding non-host species would decrease Bartonella prevalence in its principal host through reduction of host contact (encounter reduction); (2) Frequency-dependence: Adding hosts would decrease the proportion of all infected individuals in the community, resulting in a reduction in the number of contacts between susceptible and infected individuals that usually leads to transmission (transmission reduction); and (3) Dominant species effect: Dominant species, if not susceptible to Bartonellae, can constrain the abundance of susceptible hosts (susceptible host regulation). These mechanisms work in concert; and the level of Bartonella prevalence is an outcome of regulation of all of these mechanisms on the entire system.