In order to anticipate the predictable changes in the environment associated with the earth's rotation, most organisms possess intrinsic biological clocks. To be useful, such clocks require a reliable signal of 'time' from the external world. In mammals, light provides the principle source of such information; conveyed to the suprachiasmatic nucleus circadian pacemaker (SCN) either directly from the retina or indirectly via other visual structures such as the thalamic intergeniculate leaflet (IGL). Nonetheless, while the basic pathways supplying sensory information to the clock are well understood, the sensory signals they convey or how these are processed within the circadian system are not.One established view is that circadian entrainment relies on measuring the total amount of environmental illumination. In line with that view, the dense bilateral retinal input to the SCN allows for the possibility that individual neurons could average signals from across the whole visual scene. Here I test this possibility by examining responses to monocular and binocular visual stimuli in the SCN of anaesthetised mice. In fact, these experiments reveal that SCN cells provide information about (at most) irradiance within just one visual hemisphere. As a result, overall light-evoked activity across the SCN is substantially greater when light is distributed evenly across the visual scene when the same amount of light is non-uniformly distributed. Surprisingly then, acute electrophysiological responses of the SCN population do not reflect the total amount of environmental illumination. Another untested suggestion has been that the circadian system might use changes in the spectral composition of light to estimate time of day. Hence, during 'twilight', there is a relative enrichment of shortwavelength light, which is detectable as a change in colour to the dichromatic visual system of most mammals. Here I used a 'silent substitution' approach to selectively manipulate mouse cone photoreception, revealing a subset of SCN neurons that exhibit spectrally-opponent (blue-yellow) visual responses and are capable of reliably tracking sun position across the day-night transition. I then confirm the importance of this colour discrimination mechanism for circadian entrainment by demonstrating a reliable change in mouse body temperature rhythms when exposed to simulated natural photoperiods with and without simultaneous changes in colour.This identification of chromatic influences on circadian entrainment then raises important new questions such as which SCN cell types process colour signals and do these properties originate in the retina or arise via input from other visual regions? Advances in mouse genetics now offer powerful ways to address these questions. Our original method for studying colour discrimination required transgenic mice with red-shifted cone sensitivity - presenting a barrier to applying this approach alongside other genetic tools. To circumvent this issue I validated a modified approach for manipulating wildtype cone photoreception. Using this approach alongside optogenetic cell-identification I then demonstrate that the thalamic inputs to the SCN are unlikely to provide a major source of chromatic information. To further probe IGL-contributions to SCN visual responses, I next used electrical microstimulation to show that the thalamus provides inhibitory input to both colour and brightness sensitive SCN cells. Using local pharmacological inhibition I then show that thalamic inputs supress specific features of the SCN light response originating with the contralateral retina, including colour discrimination. These data thus provide new insight into the ways that arousal signals reaching the visual thalamus could modulate sensory processing in the SCN. Together then, the work described in this thesis provides important new insight into sensory control of the circadian system and the underlying neural mechanisms.