The aims in this paper are first to review theories of the evolution of bird coloration and, in some cases, partly revise and extend them, secondly to analyse the coloration of all the birds of a given geographical region using multiple regression, and thirdly on the basis of this analysis to evaluate the various theories. Theories. There have been many discussions of the selective forces acting on the coloration of birds and we review the major proposals in some detail. The earliest suggestion (Darwin 1871) was that the bright coloration of many male birds originated through sexual selection by female choice of the most exotic variants in male plumage. A conflicting view (Hingston 1933) is that brightness has an intimidatory effect on opponents, and that bright male coloration arose through sexual selection but largely due to its advantages against other males in disputes concerning mating access to females. Bird coloration has also been considered in terms of predation (see, for example, Cott 1964a). Some birds with bright plumage patterns are known to be unpalatable compared to cryptic species, and certain other patterns have been interpreted as adaptations to confuse predators. Bright colours may commonly be favoured when an individual is anyhow obvious (e.g. through activity) and where it represents an 'unprofitable' prey for a predator. This interpretation may be particularly relevant to lekking among polygamous males. A special case of unprofitable prey may be 'perception advertisement', where an animal signals (by flash patterns or alarm calls) that it has seen a predator (or opponent). It is also possible that bright coloration may serve to deflect predators away from the nest site; this requires in many ways conditions opposite to those for the unprofitable prey solution. Finally, bird colorations can act as a variety of social signals other than threat. Analysis. The coloration of the 516 species of birds that breed and/or winter in the Western Palaearctic was analysed by multiple regression. Seven regions of a bird's body plus two areas of flash coloration were recognized and scored for colour on a cryptic/conspicuous scale. Five different age/sex/season categories were recognized and scored separately for each species. These dependent variables were each analysed with respect to 17 independent variables that reflect different aspects of the reproductive biology and ecology of the birds. The advantages and disadvantages of multiple regression as an analytical technique are discussed. Results. The analysis identifies associations between the reproductive biology and ecology of the birds and the coloration of the different regions of the body of the different age/sex/season classes. Apparent exceptions to these associations are also identified and discussed. A relatively large proportion of the associations made sense in terms of the theories presented and usually there was a strong implication that for any specific association one theory was more relevant than any of the others. The results indicate that by far the most potent selective pressures to have shaped bird coloration are those related to predation risk. A number of the theories make use of predation risk, each in a different way, and for most of them some support can be gained for their involvement in the evolution of bird coloration. Of all the theories, however, it is the unprofitable prey model that seems to account for the major part of the variation in bird coloration. By contrast, no clear evidence for the involvement of sexual selection in the evolution of bird coloration could be found. Indeed, many associations, such as that between sexual dimorphism and polygamy, were more readily explicable in terms of selection pressures due to predation risk than of sexual pressures. The suggestion that bird coloration is shaped by predation rather than by sexual selection in no way prevents the coloration, as it evolves, being incorporated within the species and sex recognition system. Conclusions. It is concluded that bird coloration has evolved almost entirely in response to predation-based selective pressures. Although plumage and coloration are involved in species and sex recognition systems, they have not evolved in response to sexual selection pressures. In species that are sexually dimorphic, the male is not brightly coloured as a result of female choice or male:male competition but because he represents a less profitable prey to a predator than the females and juveniles. We predict that brightly coloured birds will most often be found to suffer less from predation than will comparable more cryptic birds (though one of the predation-risk theories does allow the converse to be true).