The number of parasites colonizing a host (termed ‘multiple infection’) is an important determinant of host-parasite interactions. In theory, multiple infection is determined by random mass action in genetically and spatially homogeneous populations of host and parasite. In real populations, deviations from these assumptions may strongly influence levels of multiple infection. We carried out inoculation experiments in microcosms of the freshwater protozoan Paramecium caudatum and its bacterial parasite Holospora undulata. Increasing parasite dose produced higher levels of (multiple) infection; more susceptible host genotypes also were more multiply infected. An overall pattern of parasite aggregation (excess of uninfected individuals and of individuals carrying larger numbers of parasites) indicated deviations from random mass-action transmission. Homogenizing spatial distributions of parasite and host in our microcosms did not affect aggregation, whereas aggregation was more pronounced in old than in new host clones. Thus, variation in susceptibility may arise over time within clonal populations. When sequentially inoculated, already established infections increased the probability of additional infection in generally resistant host clones, but decreased it in more susceptible clones. Hence, the role of multiple infection as a driver of epidemiological or evolutionary processes may vary among populations, depending on their precise genetic composition or infection history.