1. We have used the initiation of pursuit eye movements as a tool to reveal properties of motion processing in the neural pathways that provide inputs to the human pursuit system. Horizontal and vertical eye position were recorded with a magnetic search coil in six normal adults. Stimuli were provided by individual trials of ramp target motion. Analysis was restricted to the first 100 ms of eye movement, which precedes the onset of corrective feedback. 2. By recording the transient response to target motion at speeds the pursuit motor system can achieve, we investigated the visual properties of images that initiate pursuit. We have found effects of varying the retinal location, the direction, the velocity, the intensity, and the size of the stimulus. 3. Eye acceleration in the first 100 ms of pursuit depended on both the direction of target motion and the initial position of the moving target. For horizontal target motion, eye acceleration was highest if the stimulus was close to the center of the visual field and moved toward the vertical meridian. For vertical target motion, eye acceleration was highest when the stimulus moved upward or downward within the lower visual field. The shape of the relationship between eye acceleration and initial target position was similar for target velocities ranging from 1.0 to 45°/s. 4. The initiation of pursuit showed two components that had different visual properties and were expressed early and late in the first 100 ms of pursuit. In the first 20 ms, instantaneous eye acceleration was in the direction of target motion but did not depend on other visual properties of the stimulus. At later times (e.g., 80-100 ms after pursuit initiation), instantaneous eye acceleration was strongly dependent on each property we tested. Targets that started close to and moved toward the position of fixation evoked the highest eye accelerations. For high-intensity targets, eye acceleration increased steadily as target velocity increased. For low-intensity targets, eye acceleration was selective for target velocities of 30-45°/s. 5. The properties of pursuit initiation in humans, including the differences between the early and late components, are remarkably similar to those reported by Lisberger and Westbrook (12) in monkeys. Our data provide evidence that the cell populations responsible for motion processing are similar in humans and monkeys and imply that the functional organization of the visual cortex is similar in the two species.