1. An oft-cited view, derived principally from the writings of Gordon L. Walls, is that relatively few mammalian species have a capacity for colour vision. This review has evaluated that proposition in the light of recent research on colour vision and its mechanisms in mammals. 2. To yield colour vision a retina must contain two or more spectrally discrete types of photopigment. While this is a necessary condition, it is not a sufficient one. This means, in particular, that inferences about the presence of colour vision drawn from studies of photopigments, the precursors of photopigments, or from nervous system signals must be accepted with due caution. 3. Conjoint signals from rods and cones may be exploited by mammalian nervous systems to yield behavioural discriminations consistent with the formal definition of colour vision. Many mammalian retinas are relatively cone-poor, and thus there are abundant opportunities for such rod/cone interactions. Several instances were cited in which animals having (apparently) only one type of cone photopigment succeed at colour discriminations using such a mechanism. it is suggested that the exploitation of such a mechanism may not be uncommon among mammals. 4. Based on ideas drawn from natural history, Walls (1942) proposed that the receptors and photopigments necessary to support colour vision were lost during the nocturnal phase of mammalian history and then re-acquired during the subsequent mammalian radiations. Contemporary examination of photopigment genes along with the utilization of better techniques for identifying rods and cones suggest a different view, that the earliest mammals had retinas containing some cones and two types of cone photopigment. Thus the baseline mammalian colour vision is argued to be dichromacy. 5. A consideration of the broad range of mammalian niches and activity cycles suggests that many mammals are active during photic periods that would make a colour vision capacity potentially useful. 6. A systematic survey was presented that summarized the evidence for colour vision in mammals. Indications of the presence and nature of colour vision were drawn both from direct studies of colour vision and from studies of those retinal mechanisms that are most closely associated with the possession of colour vision. Information about colour vision can be adduced for species drawn from nine mammalian orders.(ABSTRACT TRUNCATED AT 400 WORDS)