BACKGROUND The normal pupillary constriction to light is an involuntary reflex that can be easily elicited and observed without specialized equipment or discomfort to the patient. Attenuation of this reflex in optic nerve disorders was first described 120 years ago. Since then, pupil examination has become a routine part of the assessment of optic nerve disease. The original cover/uncover test compares pupillomotor drive in the two eyes, but requires two working pupils and is relatively insensitive. The swinging flashlight test is now the standard clinical tool to detect pupillomotor asymmetry. It requires only one working pupil, is easily quantified, and has high sensitivity in experienced hands, but interpretation of the results needs care. Measurement of the pupil cycle time is the only clinical test that does not rely on comparison with the fellow eye, but it can only be measured in mild to moderate optic nerve dysfunction, is more time consuming, and less sensitive. Infrared video pupillography allows recordings to be made of the pupil responses to full-field or perimetric light stimulation under tightly controlled conditions with a high degree of accuracy. Frustratingly, there is a wide range in reflex gain in normal subjects limiting its usefulness unless comparison is made with the fellow eye or stimulation of unaffected adjacent areas of the visual field. In general, pupillomotor deficit shows good correlation with visual field deficit. However, some diseases of the optic nerve are associated with relative sparing either of pupil function or visual function implying that pupil tests and psychophysical tests may assess function in different subpopulations of optic nerve fibres. Less is known of the relationship between pupil measurements and electrodiagnostic tests. Pupil assessment is invaluable when distinguishing functional from organic visual loss. Its usefulness in distinguishing between different causes of optic neuropathy and as a prognostic sign is gradually emerging.