1. Horizontal and vertical eye movements were induced in normal human subjects by sinusoidal linear acceleration on a parallel swing. The swing frequency was 0.3 Hz and the peak horizontal and vertical acceleration ranged from 0.17 to 0.48 and 0.03 to 0.34 g, respectively. Eye movements were recorded with the scleral search coil technique. 2. With the subjects seated in the dark to stimulate the otolith-ocular reflex, swing displacement along the interaural axis induced horizontal eye movements with a mean sensitivity to translation (ST) (peak eye velocity/peak swing velocity) of 3.8 to 4.7 degrees/m and a mean phase shift (eye velocity re swing velocity) of -152 to -160 degrees. Vertical eye movements had ST and phase values comparable to those of the horizontal eye movements. When the subjects sat facing forward so that the horizontal linear accelerations occurred in the occipitonasal axis, almost identical vertical but no consistent horizontal eye movements were induced. In each case the horizontal and vertical eye movements were proportional to the horizontal and vertical displacement of the swing. 3. With the subject seated in the light looking to an earth-fixed target (synergistic visual-vestibular interaction), the gain (peak eye velocity/peak target velocity) of induced eye movements was near 1, and the phase was compensatory (i.e., approximately -180 degrees) for all stimuli (even at target velocities at which the pursuit gain was less than 1). Subjects were able to suppress the otolith-ocular responses by fixating on a target attached to the swing. The ST decreased by an order of magnitude compared with measurements in the dark without a fixation target. 4. Subjects were able to augment the ST (horizontal and vertical) by imagining an earth-fixed target. Halving the distance of the imagined target approximately doubled the ST. 5. In two of three subjects tested, the ST measured with mental alerting in the dark adaptively increased (approximately doubled) after 20 min of continuous synergistic visual-vestibular interaction. The subject who did not show an adaptive increase in ST began with the highest value of the 10 normal subjects. 6. We conclude that during linear accelerations of the head the otolith signal is correctly interpreted as head movement and not rotation of the gravity vector. The otolith-ocular reflex interacts with the visual pursuit system to improve ocular stability during translational head movements.