Colours appear to instantiate a number of structural properties: for instance, they stand in distinctive relations of similarity and difference, and admit of a fundamental distinction into unique and binary. Accounting for these structural properties is often taken to present a serious problem for physicalist theories of colour. This paper argues that a prominent attempt by Byrne and Hilbert (Behav Brain Sci 26:3-21, 2003) to account for the structural properties of the colours, consistent with the claim that colours are types of surface spectral reflectance, is unsuccessful. Instead, it is suggested that a better account of the structural properties of the colours is provided by a form of non-reductive physicalism about colour: a na < ve realist theory of colour, according to which colours are superficial mind-independent properties.