The possible role of spinal prostanoids in the tactile allodynia and thermal hyperalgesia associated with an experimental model of neuropathic pain was investigated. Neuropathic pain was induced by tight ligation of the L5 and L6 spinal nerves. Tactile allodynia was assessed 7 days after the surgery by measuring hindpaw withdrawal threshold to probing with von Frey filaments. Thermal hyperalgesia and nociception were determined by the 52 degrees C warm-water tail-flick test and by applying radiant heat to the plantar aspect of the hindpaw ipsilateral to the ligation. Minimal antiallodynic effect was produced by intrathecal (i.th.) administration of ketorolac or morphine up to the highest testable dose (100 microg) or by the (R)- or (S)-enantiomers of ketorolac (up to 6 microg) when administered alone. However, i.th. administration of a fixed ratio (1:1) of morphine plus racemic ketorolac or of morphine plus the (S)-enantiomer of ketorolac (S-ketorolac) produced a dose- and time-related antiallodynic effect: ED50 114 +/- 35.9 microg (total dose) for morphine plus ketorolac and 70.5 +/- 21.0 microg (total dose) for morphine plus S-ketorolac. The combination of i.th. morphine plus the (R)-enantiomer of ketorolac (R-ketorolac) (up to 200 microg total dose) was without effect. Similar antiallodynic activity was obtained for the co-administration of i.th. morphine and intravenous (i.v.) racemic ketorolac. In order to investigate the role of cyclooxygenase (COX) isozymes, relatively selective COX1 (piroxicam) and COX2 N-[2-cyclohexyloxy-4-nitrophenyl] metanesulfonamide (NS-398) inhibitors were administered i.th. (60 microg) alone or together with i.th. morphine. Piroxicam, NS-398, morphine and vehicle (90% DMSO) were without significant antiallodynic effect when administered alone, but moderate antiallodynic effects were produced by i.th. administration of fixed ratio (1:1) combinations of morphine with 60 microg each (highest soluble dose) of piroxicam (%MPE = 40.8 +/- 10.2) or NS-398 (%MPE = 32.4 +/- 9.5). Further, the combined i.th. administration of morphine, piroxicam and NS-398 in fixed 1:1:1 ratio (60 microg each) resulted in a supraadditive antiallodynic effect (%MPE = 70.4 +/- 10.8). Finally, morphine, but not ketorolac, given i.th. produced dose-dependent anti nociception in either the tail-flick or the paw-flick tests. However, there was no synergy between morphine and ketorolac against thermal nociception in either of the tests. These findings suggest that spinal prostanoids produced via both COX1 and COX2 pathways may play a role in neuropathic pain states and suggest the clinical utility of opioid plus COX-inhibitor combination therapy.