Several reports have suggested that preoperative nociceptive block may reduce postoperative pain, analgesic requirements, or both, beyond the anticipated duration of action of the analgesic agents. We have investigated, in a double-blind, placebo-controlled study, pre-emptive analgesia and the respiratory effects of preoperative administration of a multimodal antinociceptive regimen. Thirty patients undergoing thoracotomy were allocated randomly to two groups. Before surgery, the treatment group (n = 15) received morphine 0.15 mg kg-1 i.m. with perphenazine 0.03 mg kg-1 i.m. and a rectal suppository of indomethacin 100 mg, while the placebo group (n = 15) received midazolam 0.05 mg kg-1 i.m. and a placebo rectal suppository. After induction of anaesthesia, the treatment group received intercostal nerve block with 0.5% bupivacaine and adrenaline 1:200,000 (3 ml) in the interspace of the incision and in the two spaces above and two spaces below. The placebo group received identical injections but with normal saline only. The treatment group consumed significantly less morphine by patient-controlled analgesia in the first 6 h after operation, but the total dose of morphine consumed on days 2 and 3 after surgery was significantly greater in the treatment group. There were no differences between the groups in postoperative VAS scores (at rest or after movement), PaCO2 values or postoperative spirometry. However, pain thresholds to pressure applied at the side of the chest contralateral to the site of incision decreased significantly from preoperative values on days 1 and 2 after surgery in both groups. The results of this study do not support the preoperative use of this combined regimen for post-thoracotomy pain.