A new method that allows controlled masking of luminance contrast has been developed to study the use of chromatic signals in human vision. The method also makes it possible to examine the different uses of chromatic signals (e.g. the generation of perceived colour, or the construction and representation of object structure and form). By using this technique, we studied the threshold detection of chromatic signals in normal trichromats. The results show that chromatic signals are virtually unaffected by ongoing, randomly varying, luminance contrast changes. These findings suggest that chromatic signals are either processed independently or can be separated completely from any confounding luminance contrast components in the stimulus. Thresholds for detection of colour changes only, and for extraction of stimulus structure from chromatic signals in normal trichromats, in subjects with single cone receptor deficiency (i.e. dichromats) and in three subjects with abnormal colour vision caused by bilateral damage to ventromedial, extra-striate visual cortex (i.e. subjects with cerebral achromatopsia) have also been measured. No significant difference in thresholds for the two conditions was observed either in normal trichromats or in dichromats. Subjects with cerebral achromatopsia, however, reveal markedly different thresholds. The results suggest that chromatic signals are processed independently to generate perceived object colour or to construct spatially structured objects, and that these functions involve different neural substrates. The results help to explain, at least in part, why cerebral achromatopsia is a heterogeneous disorder, and why there can be significant differences in the effective use of chromatic signals in subjects described as cerebral achromatopsics.