It is generally accepted that accentuated global climatic cycles since the Plio–Pleistocene (2.8 Ma ago) have caused the intermittent fragmentation of forest regions into isolated refugia thereby providing a mechanism for speciation of tropical forest biota contained within them. However, it has been assumed that this mechanism had its greatest effect in the species rich lowland regions. Contrary evidence from molecular studies of African and South American forest birds suggests that areas of recent intensive speciation, where mostly new lineages are clustered, occur in discrete tropical montane regions, while lowland regions contain mostly old species. Two predictions arise from this finding. First, a species phylogeny of an avian group, represented in both lowland and montane habitats, should be ordered such that montane forms are represented by the most derived characters. Second, montane speciation events should predominate within the past 2.8 Ma. In order to test this model I have investigated the evolutionary history of the recently radiated African greenbuls (genus Andropadus), using a molecular approach. This analysis finds that montane species are a derived monophyletic group when compared to lowland species of the same genus and recent speciation events (within the Plio–Pleistocene) have exclusively occurred in montane regions. These data support the view that montane regions have acted as centres of speciation during recent climatic instability.