Genetic introgression from a resident species into an invading close relative can result from repeated hybridisation along the invasion front and/or allele surfing on the expansion wave. Cases where the phenomenon is massive and systematic, such as for hares (genus Lepus) in Iberia, would be best explained by recurrent hybridisation but this is difficult to prove because the donor populations are generally extinct. In the Pyrenean foothills, Lepus europaeus presumably replaced Lepus granatensis recently and the present species border is parallel to the direction of invasion, so that populations of L. granatensis in the contact zone represent proxies of existing variation before the invasion. Among three pairs of populations sampled across this border, we find less differentiation of mitochondrial DNA (mtDNA) across than along it, as predicted under recurrent hybridisation at the invasion front. Using autosomal microsatellite loci and X- and Y-linked diagnostic loci, we show that admixture across the border is quasi-absent, making it unlikely that lack of interspecific mtDNA differentiation results from ongoing gene flow. Furthermore, we find that the local species ranges are climatically contrasted, making it also unlikely that ongoing ecology-driven movement of the contact account for mtDNA introgression. The lack of mtDNA differentiation across the boundary is mostly due to sharing of mtDNA from a boreal species currently extinct in Iberia (Lepus timidus) whose mitochondria have thus remained in place since the last deglaciation despite successive invasions by two other species. Home-loving mitochondria thus witness past species distribution rather than ongoing exchanges across stabilised contact zones.