The role of 5-HT(1B/1D), 5-HT(2) and 5-HT(3) receptors in mediating descending inhibition of spinal reflexes activated by application of fentanyl to the fourth ventricle has been studied in rabbits decerebrated under N(2)O/isoflurane anaesthesia. In the control state, intraventricular fentanyl (3-30 microg kg(-1)) depressed, to an equal extent, short- and long-latency reflexes in the medial gastrocnemius muscle nerve evoked by electrical stimulation of all sural nerve afferents. Inhibition of reflexes resulted from a decreased base line excitability in the reflex pathway accompanied by a reduction in the rate of temporal summation of responses. Fentanyl-induced suppression of short- and long-latency reflexes was significantly reduced after intrathecal administration of the selective 5-HT(2)-receptor antagonist ICI 170,809 (300 microg). The same dose of the selective 5-HT(1B/1D) blocker GR 127,935 reduced inhibition from intraventricular fentanyl only for long-latency reflexes (i.e. those parts of the response for which the afferent drive is provided mainly by Adelta and C-fibre afferents). The 5-HT(3) antagonist tropisetron (also 300 microg intrathecal) did not significantly alter the descending inhibition of reflexes evoked by fentanyl. Both GR 127,935 and tropisetron reduced temporal summation of reflexes per se, effects that were reversed by intraventricular fentanyl. These data suggest that the descending pathway(s) activated by intraventricular fentanyl liberate 5-HT in the spinal cord to inhibit withdrawal reflexes by acting at 5-HT(2) and 5-HT(1B/1D), but not 5-HT(3) receptors. 5-HT(1B/1D), and to a lesser extent 5-HT(3) receptors also appear to have a role in modulating temporal summation of reflexes evoked by repetitive stimuli.