Humans identify four 'unique hues' - blue, green, yellow and red - that do not appear to contain mixtures of other colours. Unique yellow (UY) is particularly interesting because it is stable across large populations: participants reliably set a monochromatic light to a stereotypical wavelength. Individual variability in the ratio of L- and M-cones in the retina, and effects of ageing, both impact unique green (UG) settings [1,2], but cannot predict the relatively small inter-individual differences in UY [2,3]. The stability of UY may arise because it is set by the environment rather than retinal physiology. Support for this idea comes from studies of long-term, artificial chromatic adaptation [4,5], but there is no evidence for this process in natural settings. Here, we measured 67 participants in York (UK) in both the winter and summer, and found a significant seasonal change in UY settings. In comparison, Rayleigh colour matches that would not be expected to exhibit environmentally driven changes were found to be constant. The seasonal shift in UY settings is consistent with a model that reweights L- and M-cone inputs into a perceptual opponent colour channel after a small, seasonally-driven change in mean L:M cone activity.