1. Our goal was to discriminate between two classes of models for pursuit eye movements. The monkey's pursuit system and both classes of model exhibit oscillations around target velocity during tracking of ramp target motion. However, the mechanisms that determine the frequency of oscillations differ in the two classes of model. In "internal feedback" models, oscillations are controlled by internal feedback loops, and the frequency of oscillation does not depend strongly on the delay in visual feedback. In "image motion" models, oscillations are controlled by visual feedback, and the frequency of oscillation does depend on the delay in visual feedback. 2. We measured the frequency of oscillation during pursuit of ramp target motion as a function of the total delay for visual feedback. For the shortest feedback delays of approximately 70 ms, the frequency of oscillation was between 6 and 7 Hz. Increases in feedback delay caused decreases in the frequency of oscillation. The effect of increasing feedback delay was similar, whether the increases were produced naturally by dimming and decreasing the size of the tracking target or artificially with the computer. We conclude that the oscillations in eye velocity during pursuit of ramp target motion are controlled by visual inputs, as suggested by the image motion class of models. 3. Previous experiments had suggested that the visuomotor pathways for pursuit are unable to respond well to frequencies as high as the 6-7 Hz at which eye velocity oscillates in monkeys. We therefore tested the response to target vibration at an amplitude of +/- 8 degrees/s and frequencies as high as 15 Hz. For target vibration at 6 Hz, the gain of pursuit, defined as the amplitude of eye velocity divided by the amplitude of target velocity, was as high as 0.65. We conclude that the visuomotor pathways for pursuit are capable of processing image motion at high temporal frequencies. 4. The gain of pursuit was much larger when the target vibrated around a constant speed of 15 degrees/s than when it vibrated around a stationary position. This suggests that the pursuit pathways contain a switch that must be closed to allow the visuomotor pathways for pursuit to operate at their full gain. The switch apparently remains open for target vibration around a stationary position. 5. The responses to target vibration revealed a frequency at which eye velocity lagged target velocity by 180 degrees and at which one monkey showed a local peak in the gain of pursuit.(ABSTRACT TRUNCATED AT 400 WORDS)