1. The behavior of the combined eye-head gaze saccade mechanism was investigated in the rhesus monkey under both normal circumstances and in the presence of perturbations delivered to the head by a torque motor. Animals were trained to follow a target light that stepped at regular intervals through an angle of 68 degrees (+/- 34 degrees with respect to the midsagittal plane). Thus all primary saccades were center crossing. On randomly occurring trials the torque motor was pulsed so as to perturb the trajectory of the head, thus allowing us to assess both the functional state of the vestibuloocular reflex (VOR) and the effects of such perturbations on gaze saccade accuracy (gaze is defined as the sum of eye-in-head plus head-in-space, and a gaze saccade as a combined eye-head saccadic gaze shift). 2. Gaze shifts can be divided into two discrete sections: the portion during which the gaze angle is changing (the saccadic portion), and the portion during which the gaze is stationary but the head continues to move (the terminal head-movement portion). For the system to accurately acquire eccentric targets, at least two criteria must be met: 1) the saccadic portion must be accurate, and 2) the compensatory eye movement that occurs during the terminal head-movement portion must be equal and opposite to the head movement, thereby maintaining gaze stability. Perturbations delivered during the terminal head-movement portion of the gaze shift indicated that VOR was functioning normally, and thus we concluded that the compensatory eye movements that accompany head movements were vestibular in origin. 3. As reported previously, during the saccadic portion of large-amplitude gaze saccades, the VOR ceases to function. In spite of this observation, the accuracy of the gaze saccade is not affected by perturbations delivered to the head. Gaze accuracy is maintained both by changing the duration of the saccadic portion and by altering the head trajectory. 4. Because rhesus monkeys often make very rapid head movements (1,200 degrees/s), we wished to discover the velocity range over which the monkey VOR might be expected to operate. Accordingly, in a second series of experiments, VOR function was assessed during passive whole-body rotations with the head fixed. By the use of spring-assisted manual rotations, peak velocities up to 850 degrees/s were achieved. When VOR gain was measured during such rotations, it was found to be equal to 0.9 up to the maximum velocities used.(ABSTRACT TRUNCATED AT 400 WORDS)