The males of many butterflies are more brightly coloured than conspecific females and this sexual dichromatism often extends into ultraviolet (UV) wavelengths. Sexual selection, acting either in the context of intraor intersexual interactions, is thought responsible for the relative brightness of the male visage; however, this hypothesis has been difficult to test experimentally. In this study, we undertake an indirect evaluation of this idea by analysing the phylogenetic distribution of two elements of male colour pattern – UV reflectance and melanic markings – in North American sulphur butterflies. Only the UV element, which derives from nanometre-scale surface structures and is extremely bright and spectrally pure, is expected to be of sexually selected origin. We therefore predicted, based upon prior theoretical treatments, that UV colour pattern should exhibit a polyphyletic distribution typical of a sexually selected trait, whereas melanic patterns should not. We charted the distribution of each colour element across the dorsal hindand fore-wing surfaces of 26 species (spanning four genera), and summarized the resulting intraspecific variation in colour pattern using principal components analysis. Analysis of amongversus within-sister species group variance confirmed, as predicted, a largely polyphyletic distribution for UV but not melanic patterns. We also found that variation in fore-wing UV patterning could be considered orthogonal of hind-wing variation. These findings shed interesting light on the evolution of structural coloration in this group and add strength to the idea that brilliant male coloration is principally of sexually selected origin.