A 69-year-old (55 kg, 158 cm) woman was scheduled to have major gynaecological surgery due to an ovarian tumour. Before induction of anaesthesia, an epidural thoracic catheter was inserted at the T8–T9 level in a sitting position without any technical difficulty; it was tested with 4 ml of lidocaine 2% and epinephrine 1/200000 to obtain a bilateral T6-S3 sensory block. General anaesthesia was initiated thereafter with propofol, remifentanil and atracurium and maintained with desflurane and remifentanil and an infusion of atracurium for muscle relaxation under NMT monitoring (NMT Datex Ohmeda, Madison, Wisconsin, USA). The patient was installed in the supine position with both arms on abduction with a hard pillow at a low dorsal level (T10–T11) to increase surgical exposure. The surgery was carried out and lasted 6 h, around 2000 ml of blood loss was estimated and patient had a three units of red cells transfused. No haemodynamic instability was observed; mean arterial pressure was above 65 mmHg during the whole procedure. Trendelenburg position was applied from time to time because of surgical necessity. At skin closure, 20 ml of ropivacaine 0.2% was injected over a 30-min period and at the end of surgery reversal of neuromuscular blockade was performed with atropine and neostigmine under neuromuscular monitoring. The patient was extubated subsequently in the operating room after full recovery of neuromuscular blockade (90% train-of-four recovery) and a core temperature of 36°C. Soon after extubation, bilateral upper limb paraparesis was noticed (C4–C6), in association with lower limb hypoparesis. A total spinal anaesthesia was excluded because of the absence of respiratory and haemodynamic instability; therefore, this partial palsy was initially thought to be due to the spread of epidural analgesia above T4.1 However, as bilateral lower limb hypoparesis was also present, patient-controlled epidural analgesia, which consisted of voluntary boluses associated to an infusion of 3–5 ml per hour of ropivacaine, was not initiated. Two hours later, whereas lower limb paralysis faded, upper limb palsy persisted. An emergency cervical and thoracolumbar MRI was performed to exclude epidural haematoma and/or other cervical lesions. No abnormalities were detected and the following hypothesis was retained: postural bilateral upper limb palsy due to supraclavicular stretching of the brachial plexus. After 12 h, while the patient was fully awake and all vital functions were recovered and stabilised, upper limbs motricity and sensitivity was absent. Neurological consultation at day 1 postoperation confirmed our initial hypothesis; subsequently, very active physiotherapy of the upper limbs was undertaken and the patient started to recover partially both upper limb sensitivity and motricity. After 7 days, the patient could use both arms; however, the prone movement of right forehand was not possible, but she continued to improve until day 15, which she was discharged from the hospital. Neurological follow-up detected full recovery within 1 month. At 1 year, no neurological anomaly was detected. Transitional postoperative upper limb palsy is a postural complication of long surgery under general anaesthesia with muscle relaxation and special care should be given to these patients before the start of surgery by both the anaesthetic and surgical team.2 In this case, our patient had bilateral palsy, which is very uncommon. In addition, the presence of epidural analgesia initially misguided us.1,3 Nevertheless, only MRI can rule out a compression to an epidural haematoma; however, it should be performed as soon as possible (within 6 h) to eventually relieve possible compression. The injuries in our case have been previously described and are mostly due to nerve stretching under general anaesthesia, muscle relaxation and Trendelenburg position and arms on abduction and the whole situation was probably aggravated by the thoracic hard pillow that elevated the trunk. Although the outcome is generally favourable, the situation could become stressful for the patients as total nursing assistance (as in this case) would be needed at least in the early postoperative days. In the case of our patient, the concomitant presence of thoracic epidural anaesthesia was an additional confusing factor. Nevertheless, in addition to this complication, she did not benefit from postoperative epidural analgesia. Anaesthetists and surgeons should be aware of the multifactorial bases of this complication.