Objective To analyze the magnetic resonance images (MRI) of patients with spinal epidural lipomatosis (SEL) in high-altitude areas and to determine the optimal cut-off value for diagnosis with epidural fat thickness. Methods This retrospective study included patients who underwent lumbosacral MRI examination for lumbosacral pain in Ping’an District Hospital of Traditional Chinese Medicine, Haidong City, China from January 1, 2021 to December 31, 2022. The epidural fat thickness in vertebral segments T12/L1 to L5/S1 was compared between the SEL group and the non-SEL group. The diagnostic efficacy with different cut-off values at each vertebral segment was evaluated. Between-group comparisons were performed using the t-test, Mann-Whitney U test, chi-square test, or modified chi-square test. The area under the receiver operating characteristic (AUC) was used to evaluate the diagnostic efficiency. The DeLong test was used to compare AUC between the two groups. Results A total of 370 patients were included (60 in the SEL group and 310 in the non-SEL group). There were no significant differences in age, sex, height, body weight, and body mass index between the two groups (all P > 0.05). At different vertebral segments, the epidural fat thickness was significantly higher in the SEL group than in the non-SEL group (all P < 0.05). The cut-off values for SEL diagnosis with epidural fat thickness in segments T12/L1 to L5/S1 were 2.23, 4.25, 4.85, 5.57, 7.21, and 8 mm, respectively. The AUC of MRI SEL diagnosis with epidural fat thickness in segment L5/S1 was the highest (0.945, 95% confidence interval [CI]: 0.916-0.966, P < 0.001). SEL diagnosis with epidural fat thickness > 8 mm in segment L5/S1 was the most accurate, with an AUC of 0.931 (95% CI: 0.901-0.955, P < 0.001), a sensitivity of 95.0%, and a specificity of 91.3%; this AUC was significantly higher than those of diagnosis with other cut-off values (all P < 0.05). Conclusion SEL patients have significantly increased epidural fat in the spinal canal. Epidural fat thickness > 8 mm in segment L5/S1 can be used for diagnosis of SEL with improved efficiency and accuracy.