Bipolar binaural galvanic vestibular stimulation induces medium-latency (ML) emg responses in lower limb muscles and sway towards the anode ear. ML responses have a latency of 60-120ms, they are preceded by small short-latency (SL) responses of opposite polarity that have little mechanical effect. Most previous work has studied sway in the sagittal plane when the head is turned to one side. We have studied responses in the coronal plane in seven subjects facing forwards, with informed consent and ethical approval (CO1.057). Subjects were asked to stand, walk on a treadmill or bicycle using previously published methods (Ali et al. 2003; Iles et al. 2000). In standing subjects facing forward, active lower limb muscles were excited at ML in the cathode side limb and inhibited on the anode side. This was the case for posterior compartment muscles pre-activated by leaning forward (Ali et al. 2003) and anterior compartment muscles pre-activated by leaning backwards. ML responses were preceded by small SL responses of opposite polarity. Identical responses have been reported for the ankle extensors (Day et al. 1997). During slow walking the vestibular stimulus was triggered at heel strike of the investigated leg. This timing is most effective in changing the footfall of the opposite limb and deviating the path towards the anode side in free walking (Bent et al. 2004). SL and ML responses were observed in all subjects with the same pattern as during standing: in the cathode side limb ML excitation of soleus, tibialis anterior, biceps, semitendinosus, gluteus and iliacus; in the anode side limb ML inhibition. However, ML responses during walking differed in two respects from those during standing: they were very prominent in ankle muscles (doubling the emg level) and occurred at a significantly longer latency (Wilcoxon Matched Pairs Test P300 ms abolished them. During cycling vestibular stimulation was applied at top dead centre for the investigated limb. Very small SL and ML responses were observed in the cathode side limb of 2/7 subjects. These data suggest that vestibular signals have strong ML actions on the rhythmic motor output to ankle muscles during walking where balance control is essential, but not during bicycling which makes less demand on balance