Nerve ligation injury in rats may represent a useful model of some clinical neuropathic pains. Activation of N-methyl-D-aspartate (NMDA) receptors may maintain central sensitivity and contribute to neuropathic pain. Here, nerve injury was produced by unilateral ligation of the L5 and L6 spinal roots of the sciatic nerve of rats. Catheters were inserted for intrathecal (i.th.) or local delivery of drugs at the site of nerve ligation. Acute nociception was measured by the 55 degrees C water tail flick test in sham-operated and nerve-injured rats, and allodynia was determined by measuring response to von Frey filaments. In sham-operated rats, morphine (30 micrograms, i.th.) produced a 60 +/- 14.4% MPE (maximal possible effect). MK-801 pretreatment did not alter tail-flick latency or morphine antinociception in sham-operated rats. In nerve-injured rats, morphine (30 micrograms, i.th.) produced a significantly lower antinociceptive effect than in controls (34 +/- 6.3% MPE). While MK-801 alone did not alter tail-flick latency in nerve-injured rats, it significantly enhanced the antinociceptive effect of morphine to 84 +/- 16.0% MPE. Bupivacaine (0.2 ml, 0.75% w/v) at the site of injury also significantly increased the efficacy of morphine (100 +/- 0% MPE) without affecting tail flick latency alone. Bupivacaine administered at the site of injury also produced a significant antiallodynic effect of 94 +/- 7.4% MPE. The reduction in antinociceptive efficacy of i.th. morphine in nerve injured rats may be due, in part, to an ongoing spontaneous activity initiated by ectopic foci at the site of injury, and possible NMDA receptor-mediated activity of spinal neurons.