Facial electromyographic (EMG) responses were monitored intraoperatively in 67 patients with hemifacial spasm who were operated on consecutively by microvascular decompression of the facial nerve near its exit from the brain stem. At the beginning of the operation, electrical stimulation of the temporal or the zygomatic branch of the facial nerve gave rise to a burst of EMG activity (autoexcitation) and spontaneous EMG activity (spasm) that could be recorded from the mentalis muscle in all patients. In some patients, the spontaneous activity and the autoexcitation disappeared after the dura was incised or when the arachnoid was opened, but stimulation of the temporal branch of the facial nerve caused electrically recordable activity in the mentalis muscle (lateral spread) with a latency of about 10 msec that lasted until the facial nerve was decompressed in all but one patient, in whom it disappeared when the arachnoidal membrane was opened. When the facial nerve was decompressed, this lateral spread of antidromic activity disappeared totally in 44 cases, in 16 it was much reduced, and in seven it was present at the end of the operation at about the same strength as before craniectomy. In four of these last seven patients there was still very little improvement of the spasm 2 to 6 months after the operation; these four patients underwent reoperation. In two of the remaining three patients, the spasm was absent at the 3- and 7-month follow-up examination, respectively, and one had mild spasm. Of the 16 patients in whom the lateral spread response was decreased as a result of the decompression but was still present at the end of the operation, 14 had no spasm and two underwent reoperation and had mild spasm at the last examination. Of the 44 patients in whom the lateral spread response disappeared totally, 42 were free from spasm and two had occasional mild spasm at 6 and 13 months, respectively, after the operation. Monitoring of facial EMG responses is now used routinely by the authors during operations to relieve hemifacial spasm, and is performed simultaneously with monitoring of auditory function for the purpose of preserving hearing. The usefulness of monitoring both brain-stem auditory evoked potentials recorded from electrodes placed on the scalp and compound action potentials recorded directly from the eighth cranial nerve is evaluated.