The experiments were designed to explore the role of retinal inputs compared with that of the behavioral state in the modulation of the output of thalamic lateral geniculate neurons during sleep and wakefulness in cats with intact visual pathways. We made the following assumptions: the retinal dark discharge, while showing spontaneous pauses in activity, does not vary with the behavioral state; the optic tract inputs postsynaptically elicit subthreshold activities called S-potentials which in turn generate spikes, the degree of transformation being dependent on the level of alertness. On the basis of these assumptions, it could be expected that changes in retinal input frequency would modify the rate of the S-potentials. Therefore the effect of spontaneous decreases in frequency of S-potentials on the spike rate and pattern was examined in juxta- and intracellular recordings from chronically implanted cats during natural sleep and wakefulness. During quiet wakefulness and light slow-wave sleep, lateral geniculate relay neurons normally displayed numerous S-potentials associated with a moderate firing rate. Many neurons occasionally showed transient reductions in frequency of the S-potentials and an oversimplification of the discharges which combined a decreased rate with a prevalent rhythmical burst pattern. Antidromic responsiveness remained unchanged. The oscillatory periods recurred two to six times without any alteration in the control state level. They were not observed throughout wakefulness and paradoxical sleep, during which neuronal activity combined a high spike rate with a low S-potential rate. The modifications were confirmed by computation of the mean rates and of the inter-event intervals. The transfer ratio (spikes/S-potentials + spikes) significantly increased both during the oscillatory periods poor in S-potentials of quiet wakefulness and during active wakefulness. But the correlation between the transfer ratio and the spike frequency, which was high throughout the control behavioral states, faded during the periods poor in S-potentials. Thus the transient falls in the frequency of S-potentials which occurred spontaneously during quiet wakefulness caused burst discharges in lateral geniculate relay neurons, which resembled a sleep deepening, but also paralleled the effect of experimental deafferentation. The data indicate that the iterative spikes grouped in well spaced bursts which persisted during decreases in subthreshold postsynaptic activities result in an enhanced signal-to-noise ratio.(ABSTRACT TRUNCATED AT 400 WORDS)