There is considerable evidence supporting the view that the accessory optic system (AOS) and the closely associated nucleus of the optic tract (NOT) provide visual signals used in the control of optokinetic nystagmus (OKN). In frontal-eyed animals such as the cat and primate, the high degree of overlap in the visual fields of each eye, along with a substantial projection from the visual cortex, gives rise to an increased incidence of binocularly responsive neurons in the AOS. In previous studies, my collaborators and I have shown that visual cortical input to the AOS mediates ipsilateral eye responses and high speed tuning, and can function independently of the contralateral eye. However, beyond fairly gross assessments such as these, the binocular interactions of AOS cells have not been subject to detailed examination. The present study set out to determine whether the responses of binocular cells in the dorsal terminal nucleus (DTN) of the AOS are sensitive to horizontal retinal disparity. Single units were recorded from the DTN of anaesthetized, paralysed cats. A large random-dot pattern was moved under computer control at a constant velocity in the preferred and non-preferred direction. Convergent and divergent disparities were generated by deviating the visual axis of the contralateral (dominant) eye using wedge prisms. The responses of DTN units fell into three categories: (1) cells showing tuned excitatory responses (29% or 7 cells) consisting of a marked facilitation for a single or a limited range of disparities; (2) cells broadly tuned for inhibition (25% or 6 cells); and (3) cells relatively insensitive to disparity (46% or 11 cells), showing a relatively flat response profile across the entire range of disparity conditions, or cells without clear tuning. In summary, this study demonstrates that some AOS cells are sensitive to positional disparity and, therefore, this system may provide signals which specify the plane of motion for ocular stabilization. Some of these results have been presented in brief form [Grasse (1991a) Society of Neuroscience Abstracts, 17, 1380].