The prognostic value of electromyography (EMG) and its significance to estimate facial function outcome after acute facial palsy is still unclear. We retrospectively analysed the EMG reports of 494 patients with acute facial palsy treated from 1995 to 2005 in a tertiary referral centre. Initial and final facial functions were assessed by the House–Brackmann (HB) scale. Serial EMG results were classified into neurapraxia, axonotmesis/neurotmesis, mixed lesion, complete recovery, defective healing, or not classifiable. Initial HB was II–IV in 321 patients and V–VI in 173 cases. The aetiology was idiopathic palsy in 294, iatrogenic lesion in 86, traumatic in 52, Herpes zoster in 37, and of various origin in 25 patients. EMG revealed neurapraxia in 300 patients, axonotmesis/neurotmesis in 95 patients, and mixed lesion in 23 cases. EMG was not meaningful in 76 patients. The follow-up time ranged from 0.3–264 months. Final EMG revealed a full recovery in 160 patients, whereas 219 patients showed signs of defective healing. In 155 patients, EMG was not significant to classify the final outcome. The predictive EMG value for poor outcome was 77–86% and for recovery 53%. The mean EMG recovery time was 2.3 months. Mean time for defective healing was 4.3 months. Final HB was normal (HB I) in 323 patients, HB II–IV in 115 patients, and V–VI in 46 patients. We conclude that EMG has a high predictive value for unfavourable outcome after acute facial palsy. EMG is more sensible to detect signs of defective healing than clinical evaluation of facial function.