Existing theory suggests that increasing the diversity of resistance and virulence types in host–pathogen interactions will result in qualitative shifts in spatial and temporal dynamics, and greater among-population asynchrony in disease dynamics and prevalence. Here, data are presented from a biologically realistic metapopulation model of gene-for-gene interactions that indicate that population level variation in resistance diversity will be negatively associated with disease prevalence (fraction of individuals infected). The model also predicts that disease incidence (presence/absence) will be positively related to total resistance diversity across the metapopulation, because high resistance diversity also selects for more virulent pathogens. These results are then contrasted with empirical data from a natural host–pathogen system. While the argument that high resistance diversity should generally lead to lower disease levels has been applied extensively in agricultural situations, the connection between genetic diversity, resistance and disease dynamics has never been demonstrated in natural systems. Here, through analysis of multiyear data on disease prevalence in the context of knowledge of resistance variation among host populations in a natural plant host–pathogen metapopulation, the first evidence is provided that observed levels of asynchrony in disease dynamics may indeed be related to resistance structure.