Visually evoked postural responses (VEPR) to a roll-motion rotating disk were recorded from normal subjects standing on a yaw axis motorised rotating platform. The disk was fluorescent so that subjects could be tested in an otherwise dark room. Movements of the head and centre of foot pressure were measured while subjects looked at the disk with their eyes and head in the primary position and while the rotating platform moved the subjects randomly to 0, +/-45 degrees and +/-90 degrees angles from the visual stimulus. Subjects were instructed to maintain fixation on the centre of the rotating disk but the amount of horizontal eye and head movement used was not specified. Platform rotational velocity was set near threshold values for perception of self-rotation (approximately 2 degrees/s) so that subjects would find it difficult to reconstruct the angle travelled. The data showed that the VEPR occurred in the plane of disk rotation, regardless of body position with respect to the disk, and despite the subjective spatial disorientation induced by the experiment. Averages of the response revealed a good match (gain=0.95) between disk orientation and sway direction. The horizontal gaze deviation required to fixate the centre of the disk was largely achieved by head motion (head 95%, eye 5%). The results confirm previous results that VEPRs are reoriented according to horizontal gaze angle. In addition, we show that the postural reorientation is independent of cognitively or visually mediated knowledge of the geometry of the experimental conditions. In the current experiments, the main source of gaze position input required for VEPR reorientation was likely to be provided by neck afferents. The results support the notion that vision controls posture effectively at any gaze angle and that this is achieved by combining visual input with proprioceptively mediated gaze-angle signals.