1. Wind-up (frequency-dependent potentiation of the responses of spinal neurones to stimulation of unmyelinated afferents) and other N-methyl-D-aspartate (NMDA) receptor-mediated phenomena have been proposed as key mechanisms underlying persistent pain states. In this study we have compared wind-up in visceral and somatic nociceptive pathways to examine the possible contribution of these mechanisms to visceral pain and hyperalgesia. 2. Experiments were performed on thirteen decerebrate spinalized rabbits. A somato-somatic (SS) reflex (evoked by stimulating skin and muscle afferents from the L2 spinal nerve) and a viscero-somatic (VS) reflex (evoked by stimulating visceral afferents in the splanchnic nerve) were recorded from the L1 spinal nerve. The reflexes consisted of an early (A fibre) and a late (C fibre) component. 3. Conditioning trains of sixteen high intensity electrical stimuli at 1 Hz were applied to the somatic or visceral nerve. These conditioning stimuli did not produce wind-up in the early component of either reflex but evoked powerful wind-up in the late SS reflex (mean percentage of baseline +/- S.E.M., 191 +/- 30%). In contrast wind-up was weak or absent in the late VS reflex (mean percentage of baseline +/- S.E.M., 21 +/- 6%). Conditioning of somatic afferents facilitated both the early and late SS reflex but strongly depressed the early and late VS reflex. Conditioning of visceral afferents had little effect on the early SS reflex, but depressed the early VS reflex and the late components of both reflexes. 4. Intravenous administration (1-10 mg kg-1) of the NMDA receptor antagonist ketamine dose-dependently inhibited the strong wind-up in the late SS reflex and the weak wind-up in the late VS reflex, but also dose-dependently inhibited the early and late components of both baseline reflexes. 5. We conclude that neural mechanisms other than wind-up may underlie the development of visceral pain and hyperalgesia. The present results emphasize the important differences in the processing of somatic and visceral nociceptive input by spinal nociceptive systems and confirm the involvement of NMDA receptors in the spinal processing of nociceptive information.