Intravenous administration of the μ-opioid agonist, morphine, inhibits reflex responses in medial gastrocnemius (MG) motoneurones evoked by electrical stimulation of the sural nerve in decerebrated rabbits. This inhibition involves spinal and supraspinal sites of action as the effect of morphine is reduced, but not abolished, after complete spinal cord section (Lo et al. 2004). The inhibitory effect of i.v. morphine is also reduced by prior intrathecal (i.th.) administration of the selective α₂-adrenoceptor antagonist RX 821002 suggesting a role for noradrenergic bulbospinal pathways in this process. The present study has examined the mechanism of morphine-induced inhibition further by directly applying morphine supraspinally via the intracerebroventricular (i.c.v.) route. Experiments were performed on 20 rabbits decerebrated under isoflurane (3-5 %)/N₂O anaesthesia, six of which were also spinalized at L1. The left sural nerve was electrically stimulated at C fibre intensity (147 times threshold using a 0.2 ms pulse width) to evoke reflex responses in the ipsilateral MG muscle nerve. Responses were averaged, integrated by computer and analysed in 3 post-stimulus time bands: 5-12 ms (phase 1); 12-100 ms (phase 2) and 100-250 ms (phase 3). After a control period ≥ 30 min, i.c.v. morphine was given at 30 min intervals in doses of 10, 20, 70, 200 and 700 μg kg^-1 to give a total cumulative dose of 1 mg kg^-1. In 7 non-spinalized animals, 100 μg i.th. RX 821002 was given prior to i.c.v. morphine. Experiments were terminated by i.v. injection of saturated KCl solution. In decerebrated animals, i.c.v. morphine significantly (Friedman′s ANOVA, p < 0.01) inhibited all phases of the MG response so that after 1 mg kg^-1, median responses were 41%, 21% and 17% of pre-drug levels for phases 1, 2 and 3, respectively. In the presence of RX 821002, corresponding reductions were to 9%, 18% and 28% of pre-morphine controls and these decreases were significant (Friedman′s ANOVA, p < 0.001). In spinalized animals the inhibitory effect of morphine was abolished, such that after the highest dose, phase 1, 2 and 3 MG responses were a median of 118%, 111% and 87% of controls, respectively. Morphine-induced inhibition was significantly different between treatment groups for all three phases (Kruskal-Wallis ANOVA, p < 0.05) but this difference was between decerebrated versus spinalized preparations and not due to the presence of RX 821002, which in fact enhanced inhibition of short-latency responses (Dunn′s multiple comparison post-test, p < 0.05). These data show that i.c.v. morphine inhibits MG reflexes via descending bulbospinal pathways but, in contrast to i.v. morphine, this inhibition does not involve spinal α₂-adrenoceptors. Thus any interaction between morphine and noradrenergic pathways appears to be at the spinal level.