Objective: To evaluate the efficacy and safety of maximal orbital decompression surgery and intravenous glucocorticoid (ivGC) pulse therapy as the initial treatment for dysthyroid optic neuropathy (DON). Methods: Retrospective cohort study. Forty-three patients (66 eyes) with DON who were treated during April 2019 and November 2021 in Peking University People's Hospital were included. According to the treatment methods, they were divided into the glucocorticoid group (21 cases, 33eyes) and the surgery group (22 cases, 33eyes). The glucocorticoid group received a total dose of 7.5 g ivGC therapy, while the surgery group received maximal orbital decompression surgery. Clinical outcomes were recorded and assessed including pre-and post-treatment logarithm of the minimum angle of resolution best corrected visual acuity (logMAR BCVA), clinical activity score (CAS), visual evoked potential (VEP)-P100 wave amplitude and peak time, muscle index (MI), and clinical proptosis. Adverse effects were also recorded. Follow-up data at 6 months after surgery or ivGC therapy were used as the post-treatment data. Wilcoxon signed rank analysis, independent or paired sample t test, Mann-Whitney U test, Chi-square test were used for statistical analysis. Non-normal distribution data were expressed in M(IQR). Results: There was no significant difference (all P>0.05) between the two groups in terms of gender, age, course of disease or laterality, as well as baseline data such as pre-treatment BCVA, CAS, VEP-P100 wave amplitude and peak time, and MI. After treatment, logMAR BCVA of the glucocorticoid group increased significantly from 1.00 (0.65) to 0.70 (0.60), (Z=-4.55), while BCVA of the surgery group increased significantly from 1.00 (0.95) to 0.60 (0.55), (Z=-4.87). The post-treatment CAS of the glucocorticoid group decreased significantly from 5 (2) to 2 (2) (Z=-4.91), while that of the surgery group also decreased significantly from 5 (2) to 2 (2) (Z=-4.94). After treatment, the VEP-P100 wave amplitude of the glucocorticoid group increased significantly from (3.49±1.34) μV to (5.26±2.00) μV (t=-9.08), while that of the surgery group increased significantly from (3.21±1.74) μV to (4.78±2.36) μV (t=-9.24). After treatment, the P100 wave peak time of the glucocorticoid group was significantly shortened from (110.52±12.11) ms to (103.53±6.67) ms (t=4.67), while that of the surgery group was significantly shortened from (118.48±20.16) ms to (106.45±10.57) ms (t=2.84). There were significant differences between inner-group before and after treatment data (all P<0.001), whereas the inter-group after treatment data have no significant difference (all P>0.05). The orbital apex crowding relief (MI≤0.52) rate was 64% (21/33) in the glucocorticoid group and was 91% (30/33) in the surgery group, while the proportion of proptosis reduction (more than 2 mm) in the glucocorticoid group and the surgery group was 64% (21/33) and 94% (31/33), respectively. Surgery was better than ivGC therapy both in terms of orbital apex crowding relief and proptosis reduction (χ2=5.52, 7.34; both P<0.05). Serious adverse effects or complications occurred in none of the two groups. Mild adverse effects occurred in 38% (8/12) of the patients receiving glucocorticoids and in 59%(13/22) of the patients receiving surgical treatment, with no significant difference (P>0.05). Conclusions: Both maximal orbital decompression surgery and ivGC therapy can effectively improve the visual function and reduce the CAS of DON with high safety. Meanwhile, surgery has more advantages in relieving orbital apex crowding and reducing proptosis.