Ketamine and midazolam can produce analgesia following intrathecal administration in rabbits. However, neurotoxicity studies are required before these agents can be considered safe for clinical use. The aim of this study was to evaluate by histologic and blood-brain barrier (BBB) studies whether ketamine or midazolam could be used as an alternative to local anesthetics or opioids to produce spinal analgesia. Forty white New Zealand rabbits were randomly assigned to four groups of 10. In the conscious animal, 0.3 ml 0.9% saline solution, 1% lidocaine, 1% ketamine, or 0.1% midazolam was intrathecally injected intracisternally using a modification of the technique of Yaksh et al. Light and fluorescence microscopy were performed on transverse spinal cord sections by a neuropathologist unaware of the administered agents. All spinal cord section slides were scored within four zones: upper cervical, lower cervical, median thoracic, and lumbar segments. Spinal cord homogeneous lesions with higher scores than those of lidocaine-treated animals were considered abnormal. The BBB study showed evidence of neurotoxicity for ketamine, whereas light microscopy indicated no significant differences in comparison with saline and lidocaine. Midazolam-treated rabbits showed significant changes in both BBB and light microscopy studies. In view of these results, the intrathecal use of midazolam should be avoided in humans. Lesions observed following ketamine suggest the need for further experimental studies of the solvent and different ketamine enantiomers to establish definitively the safety of intrathecal free ketamine in humans.