We tested the association between host genetic diversity and resistance to a novel pathogen using controlled experiments to compare bacterially induced mortality among four populations of the endangered Gila topminnow (Poeciliopsis occidentalis) with different levels of genetic diversity, and between nine lines of inbred topminnows and their respective (non-inbred) source populations. We exposed fish in treatment groups to salinity shock and high concentrations of the bacterium Listonella (Vibrio) anguillarum, a causative agent of vibriosis in fish. Mortality associated with vibriosis (mortality) 1 to 6 days post-exposure differed among the populations, and between inbred and non-inbred samples. However, mortality was not associated with expected heterozygosity based on microsatellite loci diversity, nor was it associated with expected or observed heterozygosity at a major histocompatibility complex (MHC) class II locus. Neither was mortality related to the presence or absence of specific MHC alleles. Additionally, the effect of severe inbreeding was to reduce mortality in the inbred fish relative to the non-inbred controls in every population. Taken in the context of other studies, we conclude that effects of genetic diversity may be context- and locus-specific, environmental stress may be an important determinant of host immunity, and genetic diversity should be used cautiously to predict the potential response of a population to a specific environmental challenge.