This experiment, which extends a previous investigation (Pozzo et al. 1990), was undertaken to examine how head position is controlled during natural locomotor tasks in both normal subjects (N) and patients with bilateral vestibular deficits (V). 10 normals and 7 patients were asked to perform 4 locomotor tasks: free walking (W), walking in place (WIP), running in place (R) and hopping (H). Head and body movements were recorded with a video system which allowed a computed 3 dimensional reconstruction of selected points in the sagittal plane. In order to determine the respective contribution of visual and vestibular cues in the control of head angular position, the 2 groups of subjects were tested in the light and in darkness. In darkness, the amplitude and velocity of head rotation decreased for N subjects; these parameters increased for V subjects, especially during R and H. In darkness, compared to the light condition, the mean position of a line placed on the Frankfort plane (about 20-30 degrees below the horizontal semi-circular canal plane) was tilted downward in all conditions of movement, except during H, for N subjects. In contrast, this flexion of the head was not systematic in V subjects: the Frankfort plane could be located above or below earth horizontal. In V subjects, head rotation was not found to be compensatory for head translation and the power spectrum analysis shows that head angular displacements in the sagittal plane contain mainly low frequencies (about 0.3-0.8 Hz). The respective contribution of visual and vestibular cues in the control of the orientation and the stabilization of the head in space is discussed.