Parasite population structure is often thought to be largely shaped by that of its host. In the case of a parasite with a complex life cycle, two host species, each with their own patterns of demography and migration, spread the parasite. However, the population structure of the parasite is predicted to resemble only that of the most vagile host species. In this study, we tested this prediction in the context of a vector-transmitted parasite. We sampled the haemosporidian parasite Polychromophilus melanipherus across its European range, together with its bat fly vector Nycteribia schmidlii and its host, the bent-winged bat Miniopterus schreibersii. Based on microsatellite analyses, the wingless vector, and not the bat host, was identified as the least structured population and should therefore be considered the most vagile host. Genetic distance matrices were compared for all three species based on a mitochondrial DNA fragment. Both host and vector populations followed an isolation-by-distance pattern across the Mediterranean, but not the parasite. Mantel tests found no correlation between the parasite and either the host or vector populations. We therefore found no support for our hypothesis; the parasite population structure matched neither vector nor host. Instead, we propose a model where the parasite's gene flow is represented by the added effects of host and vector dispersal patterns.