Objective To optimize the method to elicit the facial nerve F wave and to establish its reference values and its related parameters, in order to provide an electrophysiological basis for facial nerve assessment in patients with hemifacial spasm (HFS). Methods Thirty-six healthy volunteers and 22 patients with HFS underwent this project. Compound muscle action potentials (CAMP, or M waves) and F waves were elicited by stimulating the marginal mandibular branch of the facial nerve and recorded with needle electrodes placed in the mentalis muscle. The association between F parameters and Cohen Grading of the HFS patients were analyzed. Results There were no significant differences in F parameters between men and women or between the two sides in control subjects. Minimal latency of F wave (Fmin) and mean latency of F wave (Fmean) had positive correlations with head circumference (Fmin r=0.449, P=0.013; Fmean r=0.391, P=0.033), but had no correlations with age nor height. Patients with HFS had prolonged duration of F wave (Fdura) in spasm side, compared with normal side (13.1 ms vs. 9.5 ms, P<0.01) and healthy subjects (13.1 ms vs. 9.7 ms, P<0.001), and increased ratio of F-wave amplitude and M-wave amplitude (F/M) in spasm side, compared with normal side (6.9% vs. 3.8%, P<0.001) and healthy subjects (6.9% vs. 3.7%, P<0.001). F/M exhibited a positive correlation with Cohen Grading (r=0.538, P=0.001). Conclusion A clear facial nerve F wave can be achieved by stimulating the marginal mandibular branch of the facial nerve, which provides an objective basis for evaluation of the facial nerve function in HFS patients with increased F/M and prolonged Fdura.