In comparison to mammals, relatively little is known about the ecology and phylogenetic distribution of heterothermy in birds. Reviews of avian hetereothermy have not so far revealed any clear links with climatic variables, but instead suggest heterothermy is most pronounced in phylogenetically older taxa that feed on spatially and temporally unpredictable diets such as nectar, aerial insects, or fruit. Africa’s avifauna offers rich opportunities to investigate phylogenetic and ecological correlates of heterothermy, and here I review recent work in this area. Mousebirds (Coliiformes) are a phylogenetically ancient order endemic to the Afrotropics, and have long been considered one of the avian taxa in which torpor is most pronounced. However, work on both captive and free-ranging mousebirds reveals that the ecological significance of torpor in this group needs to be viewed in the context of its interaction with communal roosting behaviour; whereas mousebirds readily enter torpor when housed individually, when roosting communally they maintain body temperatures (Tb) similar to typical avian normothermic values. Torpor is probably rare in wild mousebirds. Two other non-passerine taxa whose phylogenetic position, body mass and ecological traits suggest they might have evolved the capacity for pronounced heterothermy are barbets (Piciformes: Lybiidae) and small owls (Strigidae). Two species of barbets held in outdoor aviaries in mid-winter showed small decreases in rest-phase Tb, but not the large reductions characteristic of torpor. In a study of thermoregulation in two small owl species in the Kalahari Desert of southern Africa, free-ranging African Scops-owls (Otus senegalensis) regularly reduced skin temperature (Tskin) by up to 8.6 °C below normothermic levels, but Pearl-spotted Owlets (Glaucidium perlatum) remained normothermic throughout the study period. These findings emphasise the variation in heterothermy that can exist in related species at a single site. Patterns of heterothermy can also vary substantially within species. Among Freckled Nightjars (Caprimulgus tristigma granosus) in the more mesic eastern part of southern Africa, torpor was used frequently by some individuals, but was not correlated with lunar phase. In contrast, torpor in conspecifics (C. t. lentiginous) at a desert site was strongly linked to lunar cycles, becoming progressively deeper around new moon and shallower around full moon. These intraspecific differences likely reflect variation between sites in terms of both aridity and rainfall seasonality. In contrast to some phylogenetically older orders, passerines appear to possess a very limited capacity for torpor. Laboratory data for a small granivorous finch from the southern African arid zone supports this view, with only small decreases in Tb occurring during periods of restricted energy supply. Body temperatures approaching those associated with deep torpor have been reported in sunbirds (Nectariniidae), although the relationship between metabolic rate, Tb and air temperature in these cases suggest a modified form of shallow hypothermia, rather than classic torpor. The overall picture that emerges from studies of avian heterothermy is that many birds that a priori would seem good candidates to use pronounced torpor and/or hibernation apparently do not do so. One key factor that may select against torpor use is predation risk; the lethargic, non-responsive state characteristic of torpor likely significantly increases the risk of predation for many species. The notion that predation risk is a major evolutionary determinant of avian heterothermy is supported by recent findings that African Green-pigeons (Treron calvus) apparently do not enter torpor, whereas an ecologically similar species restricted to an island with no native predators does. Columbid birds, on account of their spectacular diversification on islands, potentially represent a very useful model taxon for testing hypotheses regarding the role of predation risk in the evolution of torpor in birds.