Although it has been reported that thyroid-stimulating immunoglobulin (TSI) is associated with the clinical characteristics of thyroid eye disease (TED), there is a paucity of literature regarding the role of TSI in diagnosing active TED. This study investigated the relationship between the level of TSI and the activity of TED and assessed the cut-off value of TSI discriminating active TED from inactive TED. This cross-sectional study included 101 patients with TED. TSI was quantitatively measured with a cell-based bioassay using a chimeric TSH receptor and a cyclic adenosine monophosphate response element-dependent luciferase. The association between TSI and a variety of demographic and clinical features of TED was analysed. Multivariate regression analysis was performed to determine possible independent factors affecting the level of TSI. TSI level was higher in males than in females (p = 0.023) and smokers than in nonsmokers (p = 0.004). TSI level was inversely correlated with the duration of ocular symptoms (r = -0.295, p = 0.003). The level of TSI was also significantly different when compared to the thyroid function (p = 0.003), TED activity (p < 0.001), and TED severity (p = 0.001). Multivariate regression analysis revealed a significant relationship between TED activity and thyroid function jointly and the TSI level. The cut-off level of TSI for predicting active TED was a specimen-to-reference ratio of 406.7 (p < 0.001, area under the curve = 0.847, sensitivity 77.4%, specificity 81.3%). TSI was a functional biomarker strongly associated with TED activity even after being adjusted by other clinical characteristics. Serum TSI level may help identify patients with active TED in clinics.