Mitochondrial DNA (mtDNA) sequencing has led to an unprecedented rise in the identification of cryptic species. However, it is widely acknowledged that nuclear DNA (nuDNA) sequence data are also necessary to properly define species boundaries. Next generation sequencing techniques provide a wealth of nuclear genomic data, which can be used to ascertain both the evolutionary history and taxonomic status of putative cryptic species. Here, we focus on the intriguing case of the butterfly Thymelicus sylvestris (Lepidoptera: Hesperiidae). We identified six deeply diverged mitochondrial lineages; three distributed all across Europe and found in sympatry, suggesting a potential case of cryptic species. We then sequenced these six lineages using double-digest restriction-site associated DNA sequencing (ddRADseq). Nuclear genomic loci contradicted mtDNA patterns and genotypes generally clustered according to geography, i.e., a pattern expected under the assumption of postglacial recolonization from different refugia. Further analyses indicated that this strong mtDNA/nuDNA discrepancy cannot be explained by incomplete lineage sorting, sex-biased asymmetries, NUMTs, natural selection, introgression or Wolbachia-mediated genetic sweeps. We suggest that this mito-nuclear discordance was caused by long periods of geographic isolation followed by range expansions, homogenizing the nuclear but not the mitochondrial genome. These results highlight T. sylvestris as a potential case of multiple despeciation and/or lineage fusion events. We finally argue, since mtDNA and nuDNA do not necessarily follow the same mechanisms of evolution, their respective evolutionary history reflects complementary aspects of past demographic and biogeographic events.