1. We recorded with one electrode action potentials of single principal cells in the lateral geniculate nucleus (l.g.n.) of cats and monkeys, together with their retinal inputs, recorded as synaptic potentials (S potentials; Bishop, Burke & Davis, 1958; Cleland, Dubin & Levick, 1971; Kaplan & Shapley, 1984). 2. We studied the effect of stimulus contrast on the transmission of visual information from the retina to the l.g.n., compared the spontaneous discharge of l.g.n. cells with that of their retinal inputs, and studied the driven (modulated) and maintained (unmodulated) discharge of l.g.n. neurones and their retinal drives. 3. The spontaneous discharge of l.g.n. cells was considerably lower than that of their retinal drives. 4. The maintained (unmodulated) discharge of l.g.n. cells during stimulation was lower than that of their retinal drives, and was largely unaffected by the stimulus contrast. 5. The responses of both the retinal input and l.g.n. cells increased with contrast, but at different rates: a given increment of contrast caused a larger increment of response in the retinal input than in the l.g.n. target cells. 6. The transmission ratio (l.g.n. response/retinal response) for most cells depended upon the stimulus contrast. This dependence indicates the presence of a non-linear contrast gain control. 7. The amount by which the l.g.n. attenuated the retinal input depended upon the temporal frequency, and, to a lesser extent, upon the spatial frequency of the stimulus. 8. The effect of contrast on signal transmission between the retina and l.g.n. was essentially the same in the macaque monkey as in the cat. 9. The attenuation of the retinal input by the l.g.n. contrast gain control could serve to prevent saturation and extend the dynamic range of cortical units, which probably receive input from several l.g.n. units.