To date, only a few reports described the potential factors influencing the position of conus medullaris. One previous study revealed no significant change of conus locations in patients with idiopathic scoliosis; however, the effect of ankylosing spondylitis (AS)-related thoracolumbar kyphosis on conus position remains unexplored. Therefore, we aimed to investigate the variation of conus medullaris terminations in patients with thoracolumbar kyphosis secondary to AS when compared with normal subjects, and evaluated the relationship between conus positions and the magnitude of kyphosis. In this study, MR images of 96 AS patients with thoracolumbar kyphosis, including 86 males and 10 females with an average of 34.6 years (range, 17–65 years), and 100 age-matched normal controls were reviewed to determine the conus terminations in relation to spinal levels. Sagittal parameters of the AS group measured on radiograph included: global kyphosis (GK), thoracic kyphosis (TK), lumbar lordosis (LL), and thoracolumbar junction (TLJ). Finally, conus tips located at the mean level of the lower 3rd of L1 in both groups, there was no significant difference of the conus distributions between AS and control group (P = 0.49). In addition, conus medullaris displayed similar positions in AS patients among various apical region groups (P = 0.88), and no significant difference was found when AS population was stratified into GK ranges of 30° (P = 0.173). Also, no remarkable correlation of the conus positions with GK (r = −0.15, P = 0.15), TK (r = −0.10, P = 0.34), LL (r = −0.10, P = 0.32), and TLJ (r = −0.06, P = 0.54) was identified. This study showed the conus terminations displayed a wide range of distributions in AS patients with thoracolumbar kyphosis, which was similar to normal subjects. Moreover, the conus located at a relatively fixed position and would not be affected by the change of kyphosis magnitude, which is an important knowledge that surgeons should acquire in surgical correction of the deformity in these patients.