Recent studies have revealed that vestibulomotor transformations contribute to maintain the hand stationary in space during trunk rotation. Here we tested whether these vestibulomotor transformations have the same latencies and whether they are subject to similar cognitive control than the visuomotor transformations during manual tracking of a visual target. We recorded hand displacement and shoulder-muscle activity in two tasks: a stabilization task in which subjects stabilized their hand during passive 30 degrees body rotations, and a tracking task in which subjects tracked with their finger a visual target as it moved 30 degrees around them. The EMG response times recorded in the stabilization task (approximately 165 ms) were twice as short as those observed for the tracking task (approximately 350 ms). Tested with the same paradigm, a deafferented subject showed EMG response times that closely matched those recorded in healthy subjects, thus, suggesting a vestibular origin of the arm movements. Providing advance information about the direction of the required arm movement reduced the response times in the tracking task (by approximately 115 ms) but had no significant effect in the stabilization task. Generally, when providing false information about movement direction in the tracking task, an EMG burst first appeared in the muscle moving the arm in the direction opposite to the actual target motion (i.e., in accord with the precueing). This behavior was rarely observed in the stabilization task. These results show that the sensorimotor transformations that move the arm relative to the trunk have shorter latencies when they originate from vestibular inputs than from visual information and that vestibulomotor transformations are more resistant to cognitive processes than visuomotor transformations.