Birds show much interspecific variation in the coloration and brightness of their plumage. I examine the hypothesis that selection due to predation on incubating birds and their nest contents can explain part of this diversity. First, I argue that rather than using absolute rates of nest predation to make predictions about the costs of conspicuous colours, we should measure experimentally whether increases in plumage conspicuousness elevate rates of nest predation. Second, I present experimental data investigating the cost of red and brown colour at ground and tree nests. These data provide the first evidence that bright colours do attract predators to nests and that, in addition, this cost varies according to the nesting site. Natural selection seems to most strongly oppose the evolution of conspicuous colours in ground-nesting birds.