Blind subjects with defective retinal processing provide a good model to study the effects of light (or absence of light) on the human circadian system. The circadian rhythms (melatonin, cortisol, timing of sleep/wake) of individuals with different degrees of light perception (n = 67) have been studied. Blind subjects with some degree of light perception (LP) mainly have normally entrained circadian rhythms, whereas subjects with no conscious light perception (NPL) are more likely to exhibit disturbed circadian rhythms. All subjects who were bilaterally enucleated showed free running melatonin and cortisol rhythms. Studies assessing the light-induced suppression of melatonin show the response to be intensity and wavelength dependent. In contrast to ocular light exposure, extraocular light failed to suppress night-time melatonin. Thus, ocular light appears to be the predominant time cue and major determinant of circadian rhythm type. Optimisation of the light for entrainment (intensity, duration, wavelength, time of administration) requires further study.