We have determined if thoracic extradural block before surgical incision for thoracotomy produces pre-emptive analgesia. Using a double-blind, placebo-controlled, crossover design, 45 patients (ASA II-III) undergoing posterolateral thoracotomy for lung resection were randomized to one of three groups: group 1 received 0.5% bupivacaine and adrenaline 1/200,000 (B+E) 8 ml through a thoracic extradural catheter (tip T3-T5) 30 min before skin incision and saline 8 ml 15 min after skin incision; group 2 received saline 8 ml extradurally before incision and B+E 8 ml after incision; group 3 received saline 8 ml extradurally before and after incision. General anaesthesia was induced and maintained with propofol, alfentanil and atracurium. The alfentanil infusion was stopped before chest closure and fentanyl 50 micrograms in saline 10 ml was given extradurally. Patient-controlled extradural analgesia (PCEA) was commenced with 0.125% bupivacaine, adrenaline 1/400,000 and fentanyl 6 micrograms ml-1 (continuous rate of 2 ml h-1 and supplementary doses of 0.5 ml per 6 min). Visual analogue scale (VAS) scores (recorded at rest, on mobilization and after cough), verbal rating scale (VRS) (recorded at rest), number of successful PCEA demands and complications were measured during the first 48 h after operation. There was no significant difference between groups, either in PCEA requirements (P > 0.21) or in VAS scores (either at rest, during mobilization of the ipsilateral arm of surgery or after cough). No significant differences between groups were found in the VRS. Thoracic extradural block with bupivacaine did not produce an early preemptive effect after thoracotomy.