How the displays of bowerbirds have evolved has attracted widespread interest. Endler et al. (2005) analyzed color use in display in a subset of bowerbird species and generalized their results to all bowerbirds. Here we discuss problems with their analysis that calls into question their conclusions. For example, they state that bowerbirds do not use decorations that match their background, but this is not supported by their results. They reconstruct historical patterns of sexual dimorphism in plumage display using questionable methodology. The high lability of these display traits makes these reconstructions unreliable and, using accepted methods and acknowledging the lability problem, we were unable to support their conclusions. Their claim that plumage differences between sympatric species are due to character displacement is not supported by the available data. Their focus is on visual contrast as the cause for display color and we offer additional hypotheses that may contribute to explaining color use. We support studies of spectral analysis of display traits but urge greater care in using this information to reach conclusions about how colorful displays have evolved.