Large reorientations of the line of sight, involving combined rotations of the eyes, head, trunk and lower extremities, are executed either as fast single-step or as slow multiple-step gaze transfers. In order to obtain more insight into the mechanisms of gaze and multisegmental movement control, we have investigated time-optimal gaze shifts (i.e. with the instruction to move as fast as possible) during voluntary whole-body rotations to remembered targets up to 180° eccentricity performed by standing healthy humans in darkness. Fast, accurate, single-step movement patterns occurred in approximately 70 % of trials, i.e. considerably more frequently than in previous studies with the instruction to turn at freely chosen speed (30 %). Head-in-space velocity in these cases was significantly higher than during multiple-step transfers and displayed a conspicuously regular bell-shaped profile, increasing smoothly to a peak and then decreasing slowly until realignment with the target. Head-in-space acceleration was on average not different during reorientations to the different target eccentricities. In contrast, head-in-space velocity increased with target eccentricity due to the longer duration of the acceleration phase implemented during trials to more distant targets. Eye saccade amplitude approached the eye-in-orbit mechanical limit and was unrelated to eye/head velocity, duration or target eccentricity. Overall, the combined movement was stereotyped such that the first two principal components accounted for data variance almost up to gaze shift end, suggesting that the three mechanical degrees of freedom under consideration (eye-in-orbit, head-on-trunk and trunk-in-space) are on average reduced to two kinematic degrees of freedom (i.e. eye, head-in-space). Synchronous EMG activity in the anterior tibial and gastrocnemius muscles preceded the onset of eye rotation. Since the magnitude and timing of peak head-in-space velocity were scaled with target eccentricity and because head-on-trunk and trunk-in-space displacements were on average linearly correlated, we propose a separate controller for head-in-space movement, whereas the movement of the eye-in-space may be, in contrast, governed by global, i.e. gaze feedback. The rapid progression of the line of sight can be sustained, and the reactivation of the vestibulo-ocular reflex would be postponed, until gaze error approaches zero only in association with a strong head-in-space neural control signal.