In rodents, acute exposure to opioids results in transient antinociception followed by longer lasting hypersensitivity to tactile or thermal stimuli, a phenomenon termed opioid-induced hyperalgesia. This hypersensitivity can be blocked or reversed by intrathecally administered cyclooxygenase inhibitors, including ketorolac, suggesting a role for spinal prostaglandins. In surgical patients, the dose of intraoperative opioid, particularly the short-acting drug, remifentanil, is directly related to increased pain and opioid requirements for many hours postoperatively. In addition, experimentally induced tactile hypersensitivity in humans is exaggerated after cessation of remifentanil infusions. The degree of this experimental opioid-induced hyperalgesia is reduced by systemic treatment with cyclooxygenase inhibitors, and investigators have speculated that this reduction reflects the actions in the central nervous system, most likely in the spinal cord. To test this hypothesis, we measured cerebrospinal fluid prostaglandin E2 concentrations during and after remifentanil infusion in 30 volunteers. These volunteers received intrathecal ketorolac or saline in a random, blinded manner during intravenous remifentanil infusion after generation of hypersensitivity by topical capsaicin. Remifentanil reduced pain to noxious heat stimuli and reduced areas of capsaicin-induced hypersensitivity similarly in those receiving intrathecal ketorolac or saline. The primary outcome measure, area of capsaicin-induced hypersensitivity after stopping remifentanil, showed a similar increase in those receiving ketorolac as in those receiving saline. Cerebrospinal fluid prostaglandin E2 concentrations did not increase during postinfusion hyperalgesia compared with those during infusion, and they were not increased during infusion compared with those in historical controls. These data fail to support the hypothesis that acute opioid-induced hyperalgesia reflects spinal cyclooxygenase activation causing central sensitization.