The genetic structure of populations is made of two components: the genetic diversity of every population (intra-population) and the genetic differentiation between every pair of populations (inter-population). These two components are influenced by genetic drift and gene flow, which are driven by the joint influence of the amount of habitat and of its spatial configuration in the landscape. Habitat amount and configuration are highly interdependent and together determine habitat connectivity, i.e. the amount of reachable habitat (ARH) at several scales. Adopting such a conception of habitat connectivity makes it possible to describe habitat patterns by considering simultaneously intra-patch and inter-patch connectivity, dispersal capacities and matrix resistance. Using an empirical genetic dataset concerning 34 large marsh grasshopper (Stethophyma grossum) populations from a Swiss agricultural landscape, we tested whether three ARH metrics computed from patch-based graphs (patch capacity, flux and betweenness centrality metrics) are good predictors of genetic structure indices computed from population-based graphs (total and private allelic richness, population-level genetic differentiation indices). The relationships between connectivity metrics and genetic indices were studied through correlation and PLS regression analyses. ARH metrics were relevant predictors of both genetic diversity and differentiation, providing an advantage over commonly used habitat metrics, such as the amount of habitat in a circular buffer around populations or the distance to the nearest habitat patch. Although in the best model allelic richness was significantly explained by three ARH metrics assessing habitat connectivity either in the focal patch or between this patch and others, genetic differentiation indices were essentially related with between-patch connectivity. Considering several matrix resistance scenarios was also key for explaining the different genetic responses. We call for a wider use of the ARH concept in future research.