Intrathecal administration of the µ-opioid receptor agonist fentanyl inhibits reflex responses of medial gastrocnemius motoneurones evoked by electrical stimulation of the sural nerve. This effect is reduced after low thoracic spinalization or intrathecal administration of the selective α2-adrenoceptor antagonist RX 821002 (Clarke et al., 1998), suggesting that activity in spinally-projecting adrenergic neurones facilitates the efficacy of spinally-applied opioids. In the present study we have investigated whether the actions of intrathecal morphine are supported in the same way.Twenty-four rabbits were decerebrated under isoflurane (2 5 %)/N₂O anaesthesia, of which 9 were spinalized at the thoracolumbar junction. Reflexes were evoked in the medial gastrocnemius (MG) muscle nerve by electrical stimulation of the ipsilateral sural nerve at an intensity sufficient to excite C fibres (134 times threshold). Responses were averaged and integrated by computer, and analyzed in 3 time bands relative to the stimulus: 5 – 12 ms (phase 1); 12 – 100 ms (phase 2) and 100 – 250 ms (phase 3). Morphine was administered intrathecally in incrementing doses of 1, 2, 7 and 20 µg kg^-1 at intervals of 30 min to give a cumulative total dose of 30 µg kg^-1.In 7 non-spinalized animals morphine was preceded by intrathecal administration of RX 821002 at 100 µg. Experiments were terminated by i.v. injection of saturated KCl solution.In decerebrated, non-spinalized rabbits morphine significantly (Friedman’s ANOVA, p < 0.04, n = 8) and non-selectively inhibited MG reflexes median values of 35, 62 and 58% of pre-morphine values after the 30 µg kg^-1 cumulative dose for phases 1, 2 and 3 respectively. In animals pre-treated with RX 821002, the equivalent changes induced by morphine were to 32, 55 and 91% of pre-morphine values, and only the changes in phase 1 and 2 reflexes were significant (Friedman’s ANOVA, p < 0.02, n = 7). In spinalized animals, morphine failed to inhibit phase 1 reflexes but reduced both phase 2 and 3 responses to median values of 48% of pre-drug values (Friedman’s ANOVA, p < 0.0001, n = 9) after the highest dose. Only the effects on phase 1 reflexes were significantly different between preparations (Kruskal-Wallis ANOVA, p < 0.01 at 30 µg kg^-1).Thus, inhibition of short-latency reflexes by spinally-applied morphine is dependent on activity in descending pathways, whereas its actions against long-latency reflexes do not appear to be. However, morphine does not interact with endogenous noradrenaline in the same way as fentanyl.