Epidural or intrathecal infusions of morphine and bupivacaine mixtures are presently used for the treatment of "refractory" cancer pain even though the neurotoxic potential of such mixtures is unknown. The pathologic findings of the spinal column, the meninges, the nerve roots, and the spinal cord, and the clinical neurologic deficits were recorded in 15 patients (5 men and 10 women), aged 26-83 (median 68) yr, treated for 4-274 (median 81) days with intrathecal infusions of morphine (with preservatives [sodium metabisulfite and sodium edetate]) and bupivacaine mixtures, given through open, subcutaneously tunneled nylon catheters. Six patients had been subjected to radiation therapy (20-96 Gy), applied over the spinal column, and four had been treated with antineoplastics believed to be neurotoxic. Ten patients had various neurologic deficits before the intrathecal treatment. The cumulative doses (ranges) given intrathecally were: morphine 33-11,900 mg, sodium metabisulfite 3.3-1,050 mg, sodium edetate 0.33-105 mg, and bupivacaine 10-41,400 mg; cumulative volumes were 16-9,400 ml. The concentrations of the drugs in the mixtures were: morphine 0.25-4.0 mg/ml, sodium metabisulfite 0.025-0.40 mg/ml, sodium edetate 0.0025-0.04 mg/ml, and bupivacaine 3.0-4.75 mg/ml. The osmolality of the mixtures in vitro ranged from 282 to 286 mOsm/kg and the pH from 4.1 to 4.6. The pathologic findings consisted of vertebral metastases (n = 6), epidural and/or intrathecal tumor masses (n = 8), focal subdural fibrosis (n = 6), infiltration of mononuclear cells in the subarachnoid space (n = 10), and discrete injuries (nerve fiber degeneration or fibrosis) to the anterior (five patients) and posterior (seven) nerve roots, and spinal cord (tumor compression [one], slight thickening of the leptomeninges [one], and thrombosis of a spinal artery and medullary infarction [one]). In none of the cases was any reaction against the nylon catheter within its subarachnoid course recorded. The neuropathologic findings were not related to the duration or cumulative doses of the intrathecal treatment. No new neurologic deficits that could be attributed to the intrathecal administration of the opiate-bupivacaine mixtures were recorded. The neuropathologic and clinical neurologic findings in cancer patients treated with intrathecal morphine-bupivacaine mixtures appeared similar to those in animals and humans reported with either intrathecal morphine or bupivacaine alone.