Eremiadinae, one of three subfamilies of Lacertidae, are distributed throughout Asia and Africa. Previous phylogenetic studies suggested that one of the main groups of Eremiadinae (the Ethiopian clade) consist of two clades with predominately East-African and South-African distribution. Yet, especially the latter one, which includes the genera Pedioplanis, Meroles, Ichnotropis, Tropidosaura and Australolacerta, was not well supported in the molecular phylogenetic analysis. In this study, we analysed the phylogenetic relationships among the genera of the ‘South African clade’ to assess whether this group actually forms a highly supported clade and to address questions concerning the monophyly of the genera. We sequenced sections of the widely used mitochondrial genes coding for 16S rRNA, 12S rRNA and cytochrome b (altogether 2045 bp) as well as the nuclear genes c-mos, RAG-1, PRLR, KIF24, EXPH5 and RAG-2 (altogether 4473 bp). The combined data set increased the support values for several nodes considerably. Yet, the relationships among five major lineages within the ‘South African clade’ are not clearly resolved even with this large data set. We interpret this as a ‘hard polytomy’ due to fast radiation within the South African lacertids. The combined tree based on nine marker genes provides strong support for the ‘South African Clade’ and its sister group relationship with the ‘East African Clade’. Our results confirm the genus Tropidosaura as a monophylum, while Ichnotropis is paraphyletic in our trees: Ichnotropis squamulosa appears more closely related to Meroles than to Ichnotropis capensis. Furthermore, the monophyly of Meroles is questionable as well. Based on our results, I. squamulosa should be transferred from Ichnotropis into the genus Meroles. Also, the two species of Australolacerta (A. australis and A. rupicola) are very distantly related and the genus is perhaps paraphyletic, too. Finally we propose a phylogeographical scenario in the context of palaeoclimatic data and compare it with a previously postulated hypothesis.