Co-evolutionary trajectories of host–parasite interactions are strongly affected by the antagonists’ evolutionary potential, which in turn depends on population sizes as well as levels of recombination, mutation, and gene flow. Under similar selection pressures, the opponent with the higher evolutionary rate is expected to lead the co-evolutionary arms race and to develop local adaptations. Here, we use mitochondrial DNA sequence data and microsatellite markers to assess the amount of genetic variability and levels of gene flow in two host–parasite systems, each consisting of an ant social parasite — the European slavemaker Harpagoxenus sublaevis and the North American slavemaker Protomognathus americanus— and its two main host species. Our population genetic analyses revealed limited gene flow between individual populations of both host and parasite species, allowing for a geographic mosaic of co-evolution. In a between-system comparison, we found less genetic variability and more pronounced structure in Europe, where previous behavioural studies demonstrated strong local adaptation. Within the European host–parasite system, the larger host species Leptothorax acervorum exhibited higher levels of both genetic variability and gene flow, and previous field data showed that it is less affected by the social parasite H. sublaevis than the smaller host Leptothorax muscorum, which has genetically depleted and isolated populations. In North America, the parasite P. americanus showed higher levels of gene flow between sites, but overall less genetic diversity than its hyper-variable main host species, Temnothorax longispinosus. Interestingly, recent ecological and chemical studies demonstrated adaptation of P. americanus to local host populations, indicating the importance of migration in co-evolutionary interactions.