Background Nucleus pulposus of intervertebral discs has proinflammatory characteristics that play a key role in neuropathic pain in lumbar herniated intervertebral disc. One of the most commonly used animal models (the traditional model) of non-compressive lumbar hemiated intervertebral disc is created by L4-L5 hemilaminectomy and the application of autologous nucleus pulposus to cover the left L4 and L5 nerve roots in rats. However, such procedures have the disadvantages of excessive trauma and low success rate. We proposed a modified model of non-compressive lumbar herniated intervertebral disc in which only the left L5 dorsal root ganglion is exposed and transplanted with autologous nucleus pulposus following incision of epineurium. We aimed to compare the modified model with the traditional one with regard to trauma and success rate.Methods Thirty Sprague-Dawley male rats were randomized into three groups: sham operation group (n=6). traditional group (n=12), and modified group (n=12). The amount of blood loss and operative time for each group were analyzed. The paw withdrawal threshold of the left hind limb to mechanical stimuli and paw withdrawal latency to heat stimuli were examined from the day before surgery to day 35 after surgery.Results Compared with the traditional group, the modified group had shorter operative time, smaller amount of blood loss, and higher success rate (91.7% versus 58.3%, P ＜0.05). There was no decrease in paw withdrawal latency in any group. The sham operation group had no decrease in postoperative paw withdrawal threshold, whereas the modified and traditional groups had significant reduction in paw withdrawal threshold after surgery (mechanical hyperalgesia).Conclusions Transplantation of nucleus pulposus onto the L5 dorsal root ganglion following incision of epineurium in rats established an improved animal model of non-compressive lumbar herniated intervertebral disc with less trauma and more stable pain ethology.