An algorithm for the identification of pathological patterns („A-trains“) of the facial nerve electromyogramm (EMG) during surgery for acoustic neuroma was developed. The quantity of A-trains was found to be predictive for postoperative functional outcome. This new method was then transferred to the operating room as a means for realtime-monitoring of A-train quantity. As a next step, the number of simultaneously analyzed EMG-channels was raised from three to nine, leading to significantly better results. The algorithm identifies a highly representative, yet small sample of overall visually identifiable A-train activity. Already with the first A-trains, a risk for facial nerve palsy results, which is low at first. With rising A-train quantity however, this risk rises to 40% quickly, and gradually continues to rise further from there. The last (and abrupt) rise of this risk with rapid approximation to 100% is observed very late with vast amounts of A-train quantity.