Central neuropathic pain occurs with multiple sclerosis, stroke, and spinal cord injury (SCI). Models of SCI are commonly used to study central neuropathic pain and are excellent at modeling gross physiological changes. Our goal was to develop a rat model of central neuropathic pain by traumatizing a discrete region of the dorsal spinal cord, thereby avoiding issues including paralysis, urinary tract infection, and autotomy. To this end, dorsal root avulsion was pursued. The model was developed by first determining the number of avulsed dorsal roots sufficient to induce below-level hindpaw mechanical allodynia. This was optimally achieved by unilateral T13 and L1 avulsion, which resulted in tissue damage confined to Lissauer's tract, dorsal horn, and dorsal columns, at the site of avulsion, with no gross physical changes at other spinal levels. Behavior following avulsion was compared to that following rhizotomy of the T13 and L1 dorsal roots, a commonly used model of neuropathic pain. Avulsion induced below-level allodynia that was more robust and enduring than that seen after rhizotomy. This, plus the lack of direct spinal cord damage associated with rhizotomy, suggests that avulsion is not synonymous with rhizotomy, and that avulsion (but not rhizotomy) is a model of central neuropathic pain. The new model described here is the first to use discrete dorsal horn damage by dorsal root avulsion to create below-level bilateral central neuropathic pain.