1. To identify vestibular influences on human walking, galvanic vestibular stimulation was applied to normal adult subjects as they walked to a previously seen target. A transmastoidal step stimulus commenced as subjects started walking. With the eyes shut, the galvanic stimulus caused large turns towards the side with the anodal current. 2. Ability to perceive the trajectory of gait without visual cues was measured by guiding blindfolded subjects from one arbitrary point to another, either walking or seated in a wheelchair. On reaching a destination position and removing the blindfold, subjects pointed to indicate the starting position. Subjects made considerable errors in estimating the trajectory, but were equally accurate whether in the wheelchair or walking. 3. To determine the effects of vestibular stimulation on the perception of trajectory, the galvanic stimulus was applied to blindfolded subjects as they were guided from one point to another in the wheelchair. The vestibular stimulus produced an illusory shift in the trajectory travelled. This shift was towards the side with the cathode, i.e. in the opposite direction to the turn produced by the stimulus during walking. 4. We conclude that galvanic vestibular stimulation during walking causes subjects to turn from their planned trajectory. In part, this altered course may compensate for an altered perception of trajectory produced by the stimulus. However, altered perception of the vertical or the base of support, or direct vestibulo-fugal influences on the leg muscles could contribute to the changes in gait.