Polymorphism in loci affecting host resistance and parasite virulence is characteristic for nearly all species and this genetic variation is considered to have profound consequences for the patterns of disease incidence, prevalence and evolution. The gene-for-gene (GFG) system is a well-characterized genetic interaction of host recognition and parasite antigenic loci for a wide range of plant-parasite interactions. Long-term maintenance of polymorphism in GFG systems has remained puzzling for both theoreticians and empiricists. Traditionally this diversity has been explained by tradeoffs with other life-history traits closely linked with fitness, yet empirical evidence for such costs has remained mixed. Here we argue that incorporating simple ecological reality – spatial structuring and gradient of environmental conditions – into host–parasite research will help us understand how polymorphism is maintained. While environmental conditions (biotic and abiotic factors) have been studied in depth in plant pathology for their influence on disease severity and plant yield, they have been rarely set into an evolutionary framework. We briefly review recent data on natural plant–parasite metapopulations and theoretical models moving from single population models towards metapopulation theory to reveal in just how many ways spatial structuring may affect the coevolutionary process. We clarify also how spatially heterogeneous selection, through G×E (or G×G×E) interactions, may be particularly important for natural host–parasite interactions and suggest that this provides the unifying ground upon which future theoretical and empirical work should be build on.