Turning the head in yaw alters the response to galvanic vestibular stimulation (GVS) in standing human subjects. The sway response is always along the inter-aural line and directed contralateral to the side of cathodal current (Lund & Broberg, 1983). This sway could be a response to either a semicircular canal signal indicating roll of the head or an otolithic signal indicating a tilt of the head, both towards the cathodal side. With the head upright and straight ahead, these movements would also apply to the whole body. If, however, the head was bent forward, the signal of head roll is a signal of body yaw, whereas the signal lateral head tilt is still a signal of lateral body tilt. The present study examines body sway and electromyographic (EMG) activity in the leg muscles evoked by GVS when subjects stand with the head at two these two different angles of head pitch; i.e. upright or bent forward.Nine subjects participated. Bilateral bipolar GVS at 2 mA was delivered as step impulses while subjects stood with the head turned to face over the left shoulder and either upright or bent forward in pitch by approximately 70 degrees. Trials were completely randomised for head position and stimulus polarity, and stimuli were delivered at random times after a “ready” signal. Antero-posterior body sway was measured at the pelvis and EMG was measured from soleus and tibialis anterior muscles.With the head upright, subjects swayed in the direction of the anodal electrode as described by Lund & Broberg, 1983. With the head bent down in pitch, this sway response was abolished. With the head level, biphasic EMG responses were as previously described by Britton et al. (1993). A short-latency response at 60-70 ms was followed by a larger medium-latency response in the opposite direction at approximately 120 ms. In soleus, the short-latency response was excitatory and the medium-latency response was inhibitory with the cathodal electrode posterior. In tibialis anterior they were in the opposite direction. With the head pitched down, the short-latency response was the same size and direction but the medium-latency response was abolished.These results are consistent with the GVS otolithic signal of body tilt being unchanged by head pitch but the semicircular canal signal changing from body roll to body yaw that does not require a postural response. Thus we can propose a model of vestibular reflex where the otolith organs and the semicircular canals evoke separate balance responses with the short-latency response from the otolith organs and the medium-latency response largely from the semicircular canals.