Repeated extinction and recolonisation events generate a landscape of host populations that vary in their time since colonisation. Within this dynamic landscape, pathogens that excel at invading recently colonised host populations are not necessarily those that perform best in host populations at or near their carrying capacity, potentially giving rise to divergent selection for pathogen traits that mediate the invasion process. Rarely, however, has this contention been empirically tested. Using Daphnia magna, we explored how differences in the colonisation history of a host population influences the invasion success of different genotypes of the pathogen Pasteuria ramosa. By partitioning the pathogen invasion process into a series of individual steps, we show that each pathogen optimises invasion differently when encountering host populations that vary in their time since colonisation. All pathogen genotypes were more likely to establish successfully in recently colonised host populations, but the production of transmission spores was typically maximised in either the subsequent growth or stationary phase of host colonisation. Integrating across the first three pathogen invasion steps (initial establishment, proliferation, secondary infection), revealed that overall pathogen invasion success (and its variance) was nonetheless highest in recently colonised host populations. However, only pathogens that were slow to kill their host were able to maximise host-facilitated dispersal. This suggest that only a subset of pathogen genotypes – the less virulent and more dispersive – are more likely to encounter newly colonised host populations at the front of a range expansion or in metapopulations with high extinction rates. Our results suggest a fundamental trade-off for a pathogen between dispersal and virulence, and evidence for higher invasion success in younger host populations; a finding with clear implications for pathogen evolution in spatiotemporally dynamic settings.