The vestibulo-ocular reflex (VOR) generates compensatory eye movements in response to angular and linear acceleration sensed by semicircular canals and otoliths respectively. Gaze stabilization demands that responses to linear acceleration be adjusted for viewing distance. This study in humans determined the transient dynamics of VOR initiation during angular and linear acceleration, modification of the VOR by viewing distance, and the effect of unilateral deafferentation. Combinations of unpredictable transient angular and linear head rotation were created by whole body yaw rotation about eccentric axes: 10 cm anterior to eyes, centered between eyes, centered between otoliths, and 20 cm posterior to eyes. Subjects viewed a target 500, 30, or 15 cm away that was extinguished immediately before rotation. There were four stimulus intensities up to a maximum peak acceleration of 2,800 degrees/s2. The normal initial VOR response began 7-10 ms after onset of head rotation. Response gain (eye velocity/head velocity) for near as compared with distant targets was increased as early as 1-11 ms after onset of eye movement; this initial effect was independent of linear acceleration. An otolith mediated effect modified VOR gain depending on both linear acceleration and target distance beginning 25-90 ms after onset of head rotation. For rotational axes anterior to the otoliths, VOR gain for the nearest target was initially higher but later became less than that for the far target. There was no gain correction for the physical separation between the eyes and otoliths. With lower acceleration, there was a nonlinear reduction in the early gain increase with close targets although later otolith-mediated effects were not affected. In subjects with unilateral vestibular deafferentation, the initial VOR was quantitatively normal for rotation toward the intact side. When rotating toward the deafferented side, VOR gain remained less than half of normal for at least the initial 55 ms when head acceleration was highest and was not modulated by target distance. After this initial high acceleration period, gain increased to a degree depending on target distance and axis eccentricity. This behavior suggests that the commissural VOR pathways are not modulated by target distance. These results suggest that the VOR is initially driven by short latency ipsilateral target distance dependent and bilateral target-distance independent canal pathways. After 25 ms, otolith inputs contribute to the target distance dependent pathway. The otolith input later grows to eventually dominate the target distance mediated effect. When otolith input is unavailable the target distance mediated canal component persists. Modulation of canal mediated responses by target distance is a nonlinear effect, most evident for high head accelerations.