The theory of "genomic islands of speciation" has been extensively debated during the last decade. This thesis not only supports this theory, but provides evidence that challenges previous beliefs on the level of admixture between species. The recently published Heliconius melpomene genome project reported apparent genomic paraphyly of H. pardalinus with regard to H. elevatus (Heliconius Genome Consortium 2012). Here, we investigate this pair of butterfly species more fully, firstly by using whole-genome resequence data, and secondly by analyzing additional geographic populations of both species, as well as outgroup taxa. Using a nuclear whole-genome phylogenetic analysis we also confirm that H. elevatus is paraphyletic. The genome-wide phylogenetic signal in H. pardalinus and H. elevatus does not indicate expected mutual monophyly of each species as it seems strongly distorted by a high level of admixture. However, several regions of the genome remain differentiated and do show the presumably original phylogenetic signal with mutual monophyly of H. pardalinus and H. elevatus. The genomic background is so homogenized that its level of differentiation (FST ~ 0.03) virtually implies panmixia. The pattern of a high level of homogenization across the genome with several regions of differentiation was consistent with a number of other statistics such as absolute divergence Dxy, nucleotide polymorphism π, number of fixed differences and with a sliding window phylogeny. The identified genomic islands of divergence comprise genes responsible for wing-patterning and chemosensation in Heliconius and some of these genes are found to be under positive selection, suggesting possible candidates of speciation.