Abstract： Background::Andersson lesions (ALs) are not uncommon in ankylosing spondylitis (AS). Plain radiography (PR) is widely used for the diagnosis of ALs. However, in our practice, there were some ALs in AS patients that could not be detected on plain radiographs. This study aimed to propose the concept of occult ALs and evaluate the prevalence and radiographic characteristics of the occult ALs in AS patients.Methods::A total of 496 consecutive AS patients were admitted in the Affiliated Drum Tower Hospital, Medical School of Nanjing University between April 2003 and November 2019 and they were retrospectively reviewed. The AS patients with ALs who met the following criteria were included for the investigation of occult ALs: (1) with pre-operative plain radiographs of the whole-spine and (2) availability of pre-operative computed tomography (CT) and/or magnetic resonance imaging (MRI) of the whole-spine. The occult ALs were defined as the ALs which were undetectable on plain radiographs but could be detected by CT and/or MRI. The extensive ALs involved the whole discovertebral junction or manifested as destructive lesions throughout the vertebral body. Independent-samples t test was used to compare the age between the patients with only occult ALs and those with only detectable ALs. Chi-square or Fisher exact test was applied to compare the types, distribution, and radiographic characteristics between detectable and occult ALs as appropriate. Results::Ninety-two AS patients with a mean age of 44.4 ± 10.1 years were included for the investigation of occult ALs. Twenty-three patients had occult ALs and the incidence was 25% (23/92). Fifteen extensive ALs were occult, and the proportion of extensive ALs was significantly higher in detectable ALs (97% vs. 44%, χ2 = 43.66, P < 0.001). As assessed by PR, the proportions of osteolytic destruction with reactive sclerosis (0 vs. 100%, χ2 = 111.00, P < 0.001), angular kyphosis of the affected discovertebral units or vertebral body (0 vs. 22%, χ2 = 8.86, P = 0.003), formation of an osseous bridge at the intervertebral space adjacent to ALs caused by the ossification of the anterior longitudinal ligament (38% vs. 86%, χ2 = 25.91, P < 0.001), and an abnormal height of the affected intervertebral space were all significantly lower in occult ALs (9% vs. 84%, χ2 = 60.41, P < 0.001). Conclusions::Occult ALs presented with more subtle radiographic changes. Occult ALs should not be neglected, especially in the case of extensive occult ALs, because the stability of the spine might be severely impaired by these lesions.