OBJECTIVE: In transcranial magnetic stimulation (TMS) of the motor cortex, the optimal orientation of the coil on the scalp is dependent on the muscle under investigation, but not yet known for facial muscles. METHODS: Using a figure-of-eight coil, we compared TMS induced motor evoked potentials (MEPs) from eight different coil orientations when recording from ipsi- and contralateral nasalis muscle. RESULTS: The MEPs from nasalis muscle revealed three components: The major ipsi- and contra-lateral middle latency responses of approximately 10 ms onset latency proved entirely dependent on voluntary pre-innervation. They were most easily obtained from a coil orientation with posterior inducing current direction, and in this respect resembled the intrinsic hand rather than the masseter muscles. Early short duration responses of around 6 ms onset latency were best elicited with an antero-lateral current direction and not pre-innervation dependent, and therefore most probably due to stimulation of the nerve roots. Late responses (>18 ms) could inconsistently be elicited with posterior coil orientations in pre-innervated condition. CONCLUSIONS: By using the appropriate coil orientation and both conditions relaxed and pre-innervated, cortically evoked MEP responses from nasalis muscle can reliably be separated from peripheral and reflex components and also from cross talk of masseter muscle activation.