Squirrel monkeys were trained to cancel their vestibulo-ocular reflex (VOR) by fixating a visual target that was head stationary during passive vestibular stimulation. The monkeys were seated on a vestibular turntable, and their heads were restrained. A small visual target (0.2 degrees) was projected from the vestibular turntable onto a tangent screen. The monkeys' ability to suppress their VOR by fixating a head stationary target while the turntable was moving was compared to their ability to pursue the target when it was moved in the same manner. Squirrel monkeys were better able to suppress their VOR when the turntable was moved at high velocities than they were able to pursue targets that were moving at high velocities. The gaze velocity gain during VOR cancellation began to decrease when the head velocity was above 80 degrees/s, and was greater than 0.6 when the head velocity was above 150 degrees/s. However, gaze velocity gain during smooth pursuit decreased significantly when the target velocity was greater than 60 degrees/s, and was less than 0.4 when the target velocity was 150 degrees/s or more. The latency of VOR suppression was significantly shorter than the latency of smooth pursuit while the monkey was cancelling its VOR. When an unpredictable step change in head acceleration was generated while the monkey was cancelling its VOR, the VOR evoked by the head acceleration step began to be suppressed shortly after the initiation of the step (approximately 30 ms). On the other hand, the latency of the smooth pursuit eye movement elicited when the visual target was accelerated in the same manner during VOR cancellation was approximately 100 ms. The comparison between these two results suggests that the monkeys did not use visual information related to target motion to suppress their VOR at an early latency. The monkeys' ability to suppress the VOR evoked by an unexpected change in head acceleration depended on the size of the head acceleration step. The VOR evoked by unexpected step changes in head acceleration was progressively less suppressed at an early latency as the size of the acceleration step increased, and was not suppressed at an early latency when the step change in head acceleration was greater than 500 degrees/s2. During smooth pursuit eye movements, unexpected step changes in head acceleration evoked a VOR that was suppressed at an early latency (approximately 50 ms) if the head movement was in the same direction as the ongoing smooth pursuit eye movement.(ABSTRACT TRUNCATED AT 400 WORDS)