There is experimental evidence of trans-synaptic retrograde degeneration of retinal ganglion cells following retrogeniculate visual pathway lesions in primate studies. Retinal nerve fibre loss in congenital homonymous hemianopia in humans is well recognized from clinical observation but the findings in acquired lesions have been controversial. Forty-eight persons were recruited and divided into three groups. Two groups were patients with retrogeniculate lesions. In the first group, the occipital damage had occurred during childhood or in adult life whilst the lesions in the second group were congenital. Inclusion criteria for the retrogeniculate lesions included: age >18 years at time of testing; homonymous hemianopia; no other ophthalmic or neurological disorder; and neuroimaging demonstration of occipital lobe damage. The third group had normal visual function. Measurement of the thickness of the peripapillary retinal nerve fibre layer by optical coherence tomography has been carried out in both eyes of each subject. The primary outcome is the peripapillary retinal nerve fibre layer thickness (RNT) in microns. The mean RNT in the eyes with temporal hemianopia (here called the 'crossing-fibre defect' eyes) is 79.8 mu (SD = 35.1 mu) in the acquired and 72.7 mu (SD = 33.2 mu) in the congenital. The mean RNT in eyes with nasal hemianopia (here called the 'non-crossing-fibre defect' eyes) is 83 mu (SD = 29.5 mu) in the acquired and 73.4 mu (SD = 26 mu) in the congenital. In the control group, the RNT measured 101.4 mu (SD = 36.6 mu) for the left eyes and 100.8 mu (SD = 35.4 mu) for the right eyes. In both crossing-fibre defect eyes and non-crossing-fibre defect eyes the mean RNT is significantly greater in the controls than in the hemianopia groups (P < 0.001). These data confirm that there is thinning of the retinal nerve fibre layer following both congenital and acquired lesions of the retrogeniculate visual pathway in humans. This is most likely to represent retinal ganglion cell loss in both congential and acquired groups. Furthermore the magnitude of the thinning is similar in both groups despite the fact that clinical observation has consistently found evidence of RNT thinning in cases of congenital but not in cases of acquired pathology. The data have also been analysed in 12 sectors around the optic disc: it has been shown that the RNT thinning follows the known trajectories of the crossing and non-crossing retinal ganglion cell axons approaching the disc.