The flora of the south-western tip of southern Africa, the Cape flora, with some 9000 species in an area of 90,000 km2 is much more speciose than can be expected from its area or latitude, and is comparable to that expected from the most diverse equatorial areas. The endemism of almost 70%, on the other hand, is comparable to that found on islands. This high endemism is accounted for by the ecological and geographical isolation of the Cape Floristic Region, but explanations for the high species richness are not so easily found. The high species richness is accentuated when its taxonomic distribution is investigated: almost half of the total species richness of the area is accounted for by 33 'Cape floral clades'. These are clades which may have initially diversified in the region, and of which at least half the species are still found in the Cape Floristic Region. Such a high contribution by a very small number of clades is typical of island floras, not of mainland floras. The start of the radiation of these clades has been dated by molecular clock techniques to between 18 million years ago (Mya) (Pelargonium) and 8 Mya (Phylica), but only six radiations have been dated to date. The fossil evidence for the dating of the radiation is shown to be largely speculative. The Cenozoic environmental history of southern Africa is reviewed in search of possible triggers for the radiations, climatic changes emerge as the most likely candidate. Due to a very poor fossil record, the climatic history has to be inferred from larger scale patterns, these suggest large-scale fluctuations between summer wet (Palaeocene, Early Miocene) and summer dry climates (Oligocene, Middle Miocene to present). The massive speciation in the Cape flora might be accounted for by the diverse limitations to gene flow (dissected landscapes, pollinator specialisation, long flowering times allowing much phenological specialisation), as well as a richly complex environment providing a diversity of selective forces (geographically variable climate, much altitude variation, different soil types, rocky terrain providing many micro-niches, and regular fires providing both intermediate disturbances, as well as different ways of surviving the fires). However, much of this is based on correlation, and there is a great need for (a) experimental testing of the proposed speciation mechanisms, (b) more molecular clock estimates of the age and pattern of the radiations, and (c) more fossil evidence bearing on the past climates.