A retrospective analysis of 413 patients who received postoperative epidural analgesia under a standardized protocol found that 84 (20%) had a duration of epidural catheterization of greater than four days. The most common reasons were significant pain (n=64, 15%) and coagulopathy (n=26, 6%). Risk factor analysis for coagulopathy showed an odds ratio of 10.1 (95% confidence interval 4.2-24.5) for prolonged epidural catheterization among patients undergoing hepatectomy. Magnetic resonance imaging, performed in four patients with clinical signs suggestive of epidural haematoma, was negative for a space-occupying lesion in all cases. Eleven patients developed fever and clinical signs suggestive of epidural catheter-related infection, necessitating early catheter removal. Sixteen patients had persistent lower limb weakness at 24 hours after catheter removal. The signs soon resolved in all except two, one of whom had neuropathy related to intraoperative positioning and the other preoperative weakness. Accidental epidural catheter dislodgement occurred in 29 patients (7%) and is potentially hazardous if coagulopathy is unresolved. The risk-benefit ratio and factors complicating catheter removal, especially coagulopathy, should be considered when deciding whether to use epidural techniques.