Upper abdominal and thoracic surgeries require efficient pain management. The complications of postoperative analgesia include respiratory depression and--when choosing the epidural route--possible damage to the spinal cord by infection, trauma, or bleeding. Therefore, thoracic epidural analgesia may appear to be too risky and is frequently cancelled although many studies have shown its excellent efficacy. Controlled studies comparing thoracic epidural analgesia to lumbar epidural analgesia or intravenous analgetic regimens with special regard to the patient's outcome are contradictory. To make the preoperative decision on the method of pain control more rational, we studied catheter-related complications from 2056 thoracic epidural catheters used for intra- and postoperative analgesia retrospectively (n = 1002) and prospectively (n = 1054) over a 5 1/2-year period. In all patients the thoracic epidural catheter was inserted preoperatively using local anaesthesia, in most cases by the paramedian approach between level T 5/6 and T 8/9. During the clinical course of all patients there were no clinical signs of any epidural bleeding or infection. Neurological complications caused by the epidural catheter did not occur. Seven patients (0.035%) experienced radicular pain that disappeared after removal of the catheter or interruption of the puncture, respectively. A primary perforation of the dura mater was noticed in 0.5% of cases retrospectively and 1.23% prospectively. Respiratory depression following epidural application of 0.3 mg buprenorphine was seen in 1 patient (0.05%). Continuous analgesia with local anaesthetics and/or opioids applied epidurally by a thoracic catheter was performed on the peripheral ward (n = 829, 40%) if close monitoring of the neurological status as well as rapid diagnosis of any painful paraesthesia or paraplegia was possible.