Posterior spinal fusion for correction of scoliosis is a major procedure for which the provision of satisfactory, safe postoperative analgesia is often a problem. One possible solution involves the placement of epidural catheters under direct vision by the surgeon at the end of the procedure, followed by an epidural infusion of local anaesthetic with or without an opioid. Despite its simplicity, this technique has not been reported as being consistently successful. We report an observational study of the analgesia achieved with surgically placed epidural catheters and of the reasons for the failure of the technique. Fourteen consecutive patients undergoing posterior spinal fusion had epidural catheters placed by the surgeon and had radio-opaque dye injected down the catheter 15 min before their routine postoperative chest X-ray. Analgesia was assessed at 0, 6, 12 and 24 h after surgery using visual analogue scores. Five patients had inadequate pain control; none of these patients had dye visible in the epidural space. Seven patients had dye visible in the epidural space; all of these cases had satisfactory analgesia. In two cases, dye was observed in the paravertebral gutters; both of these patients had satisfactory postoperative analgesia. This small pilot study suggests that correctly placed 'surgical' epidural catheters are capable of providing good analgesia after posterior spinal fusion and that misplaced catheters, seen in a large proportion of patients, are associated with inadequate analgesia.