Abstract The post-Kibaran sedimentary rocks of the Natal Group have traditionally been regarded as the lateral equivalents of the Palaeozoic Cape Supergroup. However, the lithology in Transkei and southern Natal (pale grey marine quartz arenites with late Devonian fossils) is markedly different from that of the rest of Natal (unfossiliferous fluviatile red-beds, with a newly recognised volcanic component). These two lithofacies do not interdigitate as previously thought; there is a break in outcrop over the Dweshula Basement High, with quartz arenites cropping out to the south and red-beds to the north. There is thus no reason to relate the two “facies” in space or time and the term “Natal Group” should be restricted to the northern red beds. The southern rocks (Msikaba Formation) are probably lateral equivalents of the Witteberg Group. Petrographic, KAr and 40Ar/ 39Ar isotopic data from samples of the basal Durban Formation (Natal Group) reveal a complex history of isotopic resetting. Pan-African KAr ages of ∼580 Ma were obtained from detrital muscovites, whilst secondary clay fractions suggest major KAr components at ∼400, ∼350 and ∼260 Ma. The last event reflects partial overprinting during the Cape Orogeny, but the significance of the older dates is less certain. 40Ar/ 39Ar spectrum analyses of two samples suggest that the Natal Group was deposited at ∼490 Ma (Arenig). Thus, the rocks probably represent a fluviatile molasse deposit, derived from the rapid erosion of a Pan-African mountain chain situated to the northeast, in present-day Moçambique, and laid down in an adjacent fault-bounded trough during the early Ordovician. This conclusion is supported by palaeocurrent data.