BACKGROUND: Morphine has a direct action on morphine receptors in the brain and spinal cord. Intrathecally administered L-NAME, a nitric oxide synthase inhibitor, is known to have an antinociceptive effect on formalin-induced pain in animal studies. Efficacy of intrathecally administered ketorolac, a cyclooxygenase inhibitor, is somewhat controversial. The interactions of intrathecally administered morphine, ketorolac and L-NAME on formalin-induced nociception was studied. METHODS: Male Sprague-Dawley rats were implanted with chronic lumbar intrathecal catheters and were tested for paw flinch by a formalin injection. Drugs were intrathecally administered 15 min before the formalin injection, and biphasic painful behaviors were observed. We obtained the ED50 for each agent (ketorolac, L-NAME and morphine). ED50 fractions (1, 1/2 and 1/4) of drug combinations of L-NAME-ketorolac, morphine-L-NAME and ketorolac-morphine were administered. The ED50 of each combined drug was established and isobolographic analysis of the drug interactions was carried out. RESULTS: Intrathecal administration of ketorolac, L-NAME and morphine produced a dose-dependent suppression of pain behaviors in phase 2. ED50 values were 297.04micro gram for ketorolac, 207.46micro gram for L-NAME and 0.17micro gram for morphine in phase 2. Isobolographic analysis showed that the combination of intrathecal morphine and L-NAME synergistically reduced pain behaviors in phase 2. CONCLUSIONS: Intrathecally administered morphine, L-NAME and ketorolac produced a dose-dependent decrease in the number of paw flinches in both phase 1 and phase 2 on the formalin test. Morphine with L-NAME showed synergistic analgesic effects on formalin-induced pain in phase 2.