There are clinical observations that neurogenic pain can respond well to the opioid ketobemidone, in contrast to pethidine and morphine. This has led us to the hypothesis that the analgesic effect of ketobemidone in neurogenic pain may be due to both opioid as well as additional non-opioid effects. The present study was therefore made to evaluate the effects of ketobemidone. The study consists of two parts. (1) Single unit recordings were made from dorsal horn neurones in the halothane-anaesthetised rat. Neurones were activated by transcutaneous electrical stimulation of their receptive fields at C-fibre strength and their responses quantified. The wind-up of the neurones, due to N-methyl-D-aspartate (NMDA) receptor activation, leading to marked increases in C-fibre responses and an associated post-discharge was also measured. Ketobemidone, applied to the spinal cord, equivalent to an intrathecal injection, dose-dependently and selectively reduced C-fibre evoked responses. Ketobemidone was also found to block wind-up more effectively than morphine at equieffective doses, but unlike morphine in a non-naloxone-reversible manner. (2) In a binding study ketobemidone was shown to inhibit [3H]MK-801 binding with a Ki value of 26 microM. Therefore, ketobemidone appears to possess both mu opioid agonist as well as NMDA blocking effects.