The population structure of parasites is central to the ecology and evolution of host-parasite systems. Here, we investigate the population genetics of Pasteuria ramosa, a bacterial parasite of Daphnia. We used natural P. ramosa spore banks from the sediments of two geographically well-separated ponds to experimentally infect a panel of Daphnia magna host clones whose resistance phenotypes were previously known. In this way, we were able to assess the population structure of P. ramosa based on geography, host resistance phenotype and host genotype. Overall, genetic diversity of P. ramosa was high, and nearly all infected D. magna hosted more than one parasite haplotype. On the basis of the observation of recombinant haplotypes and relatively low levels of linkage disequilibrium, we conclude that P. ramosa engages in substantial recombination. Isolates were strongly differentiated by pond, indicating that gene flow is spatially restricted. Pasteuria ramosa isolates within one pond were segregated completely based on the resistance phenotype of the host—a result that, to our knowledge, has not been previously reported for a nonhuman parasite. To assess the comparability of experimental infections with natural P. ramosa isolates, we examined the population structure of naturally infected D. magna native to one of the two source ponds. We found that experimental and natural infections of the same host resistance phenotype from the same source pond were indistinguishable, indicating that experimental infections provide a means to representatively sample the diversity of P. ramosa while reducing the sampling bias often associated with studies of parasite epidemics. These results expand our knowledge of this model parasite, provide important context for the large existing body of research on this system and will guide the design of future studies of this host-parasite system.