We performed a double-blinded, randomized, controlled trial in 15 patients to determine the efficacy of intrathecal morphine or clonidine, alone or combined, in the treatment of neuropathic pain after spinal cord injury. The combination of morphine and clonidine produced significantly more pain relief than placebo 4 h after administration; either morphine or clonidine alone did not produce as much pain relief. In addition, lumbar and cervical cerebrospinal fluid (CSF) concentrations, sampled at these levels at different times after administration were examined for a relationship between pain relief and CSF drug concentration. Lumbar CSF drug concentrations were initially several orders of magnitude larger than those in cervical CSF. After 1-2 h, the concentrations of morphine in cervical CSF markedly exceeded those of clonidine. The concentration of morphine in the cervical CSF and the degree of pain relief correlated significantly. We conclude that intrathecal administration of a mixture of clonidine and morphine is more effective than either drug administered alone and is related to the CSF-borne drug concentration above the level of spinal cord injury. If there is pathology that may restrict CSF flow, consideration should be given to intrathecal administration above the level of spinal cord damage to provide an adequate drug concentration in this region.