In a quality improvement audit on epidural analgesia in 300 patients after major abdominal surgery, we identified postoperative lower leg weakness and premature catheter dislodgement as the most frequent causes of premature discontinuation of postoperative epidural infusion. Lower limb motor weakness occurred in more than halfofthe patients with lumbar epidural analgesia. In a second period monitoring 177 patients, lumbar catheter insertion was abandoned in favour of exclusive thoracic placement for epidural catheters. Additionally, to prevent outward movement, the catheters were inserted deeper into the epidural space (mean (SD) 5.2 (1.5) cm in Period Two vs 4.6 (1.3) cm in Period One). Lower leg motor weakness declined from 14.7% to 5.1% (odds ratio 0.35; 95% confidence interval 0.16-0.74) between the two periods. Similarly, the frequency of premature catheter dislodgement was reduced from 14.5% to 5.7% (odds ratio 0.35; 95% confidence interval 0.17-0.72). With a stepwise logistic regression model we demonstrated that the odds of premature catheter dislodgement was reduced by 43% for each centimetre of additional catheter advancement in Period Two. We conclude that careful audit of specific complications can usefully guide changes in practice that improve success of epidural analgesia regimens.