The distribution of human corticobulbar motor excitatory and inhibitory output is not fully understood. In particular, it is unclear whether the pattern of innervation is the same for upper and lower facial muscles, and what is the motor cortical area giving rise to such innervation. We used electrodes implanted in the subthalamic nucleus (STN) in patients with Parkinson's disease to activate motor tracts at a subcortical level. We examined the excitatory and inhibitory effects of unilateral single STN deep brain stimulation (sSTN-DBS) in 14 patients by taking recordings from facial, cervical and upper limb muscles on both sides. We measured the latency and amplitude of the motor-evoked potentials (MEPs), and the latency and duration of the silent periods, and compared ipsilateral with contralateral responses and responses obtained in different muscles. Unilateral sSTN-DBS induced strictly contralateral MEPs in the trapezius, deltoid, biceps and thenar muscles. The same stimulus always induced bilateral MEPs in the orbicularis oculi, orbicularis oris, masseter and sternocleidomastoid at a mean latency in the range 6.0-9.1 ms. MEP latencies in the orbicularis oculi and orbicularis oris were significantly longer than in the masseter and sternocleidomastoid (P < 0.01). A short latency small action potential was recorded in the ipsilateral orbicularis oculi that was likely generated by activation of extraocular muscles. During sustained voluntary muscle contraction, a silent period was recorded at similar onset latency on both sides. This period was significantly shorter in orbicularis oculi than in masseter, and in the ipsilateral side for both muscles (P < 0.01). sSTN-DBS is able to activate the descending projecting fibres in the corticobulbar tract eliciting bilateral MEPs and silent periods in facial and cranial muscles. This suggests that fibres to both ipsi- and contralateral motor nuclei descend together at the level of the STN. These findings are relevant in the discussion of the innervation of upper and lower facial muscles in humans and in the interpretation of previous results obtained with transcranial cortical stimulation.