The present study investigated the regional distribution of the N-methyl-D-aspartate (NMDA) receptor containing the NR2B subunit protein in rat lumbar spinal cord and examined whether selective NR2B antagonists would exhibit antinociception with reduced side-effect liability than subtype non-selective NMDA antagonists and anticonvulsants. Immunocytochemical studies showed the NR2B subunit had a restricted distribution, with moderate labelling of fibres in laminas I and II of the dorsal horn suggesting a presynaptic location on primary afferent fibers and possible involvement in pain transmission. In the in vivo studies, the NMDA/glycine antagonists (MK-801, 0.02-1 mg/kg i.p., L-687,414 10-300 mg/kg i.p., and L-701,324 1-10 mg/kg i.p.) and the anticonvulsant, gabapentin (10-500 mg/kg p.o.), induced rotarod deficits at antinociceptive doses. In contrast, the selective NR2B antagonists, (+/-)-CP-101,606 (1-100 mg/kg p.o.) and (+/-)-Ro 25-6981 (3-100 mg/kg i.p.) showed a significant dose window. (+/-)-CP-101,606 caused no motor impairment or stimulation in rats at doses up to 100 mg/kg p.o., which is far in excess of those inhibiting allodynia in neuropathic rats (ID50 4.1 mg/kg, p.o.). (+/-)-Ro 25-6981 also showed a significant separation (ID50 allodynia 3.8 mg/kg, i.p.), however, some disruption of rotarod performance was observed at 100 mg/kg. The anticonvulsant lamotrigine (3-500 mg/kg p.o.) also showed a good dose window. These findings demonstrate that NR2B antagonists may have clinical utility for the treatment of neuropathic and other pain conditions in man with a reduced side-effect profile than existing NMDA antagonists.