Diurnal dwarf geckos of the genus Lygodactylus are distributed in tropical and subtropical regions and live in highly diverse habitats. The genus currently comprises 79 species and several candidates for new species or subspecies. Most of these taxa occur in Sub-Saharan Africa and Madagascar, with only two described species in South America. Although the main center of diversity of Lygodactylus currently is Africa, the genus probably has a Malagasy origin, followed by two or three independent transoceanic dispersal events between Madagascar and Africa and one trans-Atlantic dispersal from Africa to South America. A few species colonised islands in the Western Indian Ocean belonging to the Zanzibar Archipelago and to the Îles Éparses. Here we examined L. grotei pakenhami from Pemba Island, L. insularis from Juan de Nova, and L. verticillatus from Europa Island to clarify their taxonomic status and their origin. Concerning L. grotei pakenhami and L. insularis, preceding studies pointed to a relation to species of the African L. capensis group. In contrast, L. verticillatus on Europa Island is considered to be conspecific with Malagasy populations. Therefore, we conducted a phylogenetic study of the African L. capensis group and the Malagasy L. verticillatus group, and examined color pattern, selected morphological characters and two mitochondrial markers (ND2 for African and 16S rRNA for Malagasy Lygodactylus). Lygodactylus grotei pakenhami from Pemba and L. grotei from mainland Africa cannot be distinguished by their scalation, but their reciprocal monophyly suggested by mitochondrial DNA, conspicuously different coloration (both in adults and hatchlings) and their high genetic distances (16.3% in ND2) support the hypothesis that these taxa represent two distinct species. Consequently, we elevate L. grotei pakenhami to species level, as Lygodactylus pakenhami Loveridge, 1941. Lygodactylus pakenhami is endemic to Pemba Island which was possibly separated from the African mainland during the late Miocene or Early Pliocene (6 million years ago). The simplest explanation for the existence of L. pakenhami on Pemba is vicariance. A recent, human-mediated transportation is excluded, as the molecular data clearly indicate a longer period of isolation. Lygodactylus insularis has been supposed to be related to the taxa 'capensis' or 'grotei'. However, it is impossible to discern the relationship of L. insularis, L. capensis and L. grotei by means of scalation or coloration alone. Our molecular phylogenetic analyses reveal that L. insularis is embedded within the L. capensis group, clearly indicating its African origin. The single gene (ND2) as well as the multigene analyses fully support a closer common origin of L. insularis and L. capensis than of L. insularis and L. grotei. However, the position of L. insularis within the clade formed by L. insularis, L. nyaneka, L. capensis sensu stricto and six L. aff. capensis groups is not clearly resolved. Lygodactylus insularis is endemic on Juan de Nova Island, an old low elevation atoll. That all L. insularis mitochondrial sequences are very similar to each other and together form a monophyletic lineage is in agreement with the hypothesis of a single dispersal event to the island. For the L. verticillatus population from Europa Island our mitochondrial data suggest close relationships to conspecific samples from the coastal regions of south-western Madagascar. As we found no relevant morphological or genetic differences between the insular and the Malagasy populations of L. verticillatus, and no remarkable genetic variation within the monophyletic lineage on Europa, we suggest a single, very recent dispersal event, perhaps human-mediated. Although the genus Lygodactylus colonised Africa, islands in the Gulf of Guinea, South America and some islands in the Western Indian Ocean, it seems-compared to other lizard genera-to be only moderately successful in transoceanic long-distance dispersal.