The discharge of neurons in the vestibular nuclei was recorded in alert squirrel monkeys while they were being sinusoidally rotated at 2 Hz. Type I position-vestibular-pause (PVP I) and vestibular-only (V I) neurons, as well as a smaller number of other type I and type II eye-plus-vestibular neurons were studied. Many of the neurons were monosynaptically related to the ipsilateral vestibular nerve. Eye-position and vestibular components of the rotation response were separated by multiple regression. Anodal currents, simultaneously delivered to both ears, were used to eliminate the head-rotation signals of irregularly discharging (I) vestibular-nerve afferents, presumably without affecting the corresponding signals of regularly discharging (R) afferents. R and I inputs to individual central neurons were determined by comparing rotation responses with and without the anodal currents. The bilateral currents, while reducing the background discharge of all types of neurons, did not affect the mean vestibular gain or phase calculated from a population of PVP I neurons or from a mixed population consisting of all type I units. From this result, it is concluded that I inputs are canceled at the level of secondary neurons. The cancellation may explain why the ablating currents do not affect the gain and phase of the vestibulo-ocular reflex. While cancellation was nearly perfect on a population basis, it was less so in individual neurons. For some neurons, the ablating currents decreased vestibular gain, while for other neurons the vestibular gain was increased. The former neurons are interpreted as receiving a net excitatory (I-EXC) I input, the latter neurons, a net inhibitory (I-INH) input. When compared with the corresponding R inputs, the I inputs were usually small and phase advanced. Phase advances were larger for I-EXC than for I-INH inputs. The sign and magnitude of the I inputs were unrelated to other discharge properties of individual neurons, including discharge regularity and the phase of vestibular responses measured in the absence of the ablating currents. Unilateral currents were used to assess the efficacy of ipsilateral and contralateral pathways. Ipsilateral pathways were responsible for almost all of the effects seen with bilateral currents. The results suggest that the vestibular signals carried by central neurons, even by those neurons receiving a monosynaptic vestibular-nerve input, are modified by polysynaptic pathways.