OBJECTIVES: The epithelial cell adhesion molecule (EpCAM) is a transmembrane glycoprotein that is overexpressed in various epithelial tumors and has been associated with tumor aggressiveness and unfavorable prognosis. This study aimed to explore the association between ultrasonic characteristics and EpCAM expression in papillary thyroid carcinoma (PTC), as well as the value in predicting the risk of cervical lymph node metastasis (CLNM). METHODS: Participants with suspected thyroid cancer underwent conventional and contrast-enhanced ultrasonography (CEUS) before surgery. Age, sex, and nodule features in ultrasound were recorded. Participants were classified based on the presence of CLNM. EpCAM expression was detected using immunohistochemistry. The risk predictors of CLNM were evaluated by using univariate and multivariate Cox model. RESULTS: The final dataset included 53 participants, of whom 27(50.94%) had CLNM, while 26(49.06%) did not. The EpCAM-positive area (p < 0.001), score (p < 0.001), and nuclear expression (p = 0.027) were significantly different between the CLNM and non-CLNM groups. Receiver operating characteristic curve analysis showed that EpCAM score with 4.5 had sensitivity and specificity of 85.2% and 76.9%, respectively, for diagnosing CLNM. Multivariate logistic regression results showed that overexpression EpCAM (score ≧ 4.5) [OR:46.54, p = 0.007], tumor size ≧ 1.05 cm [OR:14.21,p = 0.01] and extrathyroidal extension (ETE) on CEUS [OR:113.33, p = 0.01] were independent risk factors of CLNM. Overexpression of EpCAM (62.07%, 18/29) correlated with ETE of PTC confirmed by postoperative pathology (p = 0.01, 0.027). No significant difference in PTC histological variants was detected based on EpCAM expression (p = 0.05). CONCLUSIONS: EpCAM overexpression demonstrated a significant association with ETE and CLNM in PTC. The integration of ultrasonography with EpCAM expression profiling offers A promising multiparameter diagnostic model to enhance predictive accuracy. Further studies evaluating EpCAM expression in fine-needle aspiration biopsy (FNAB) samples are required to determine its possible preoperative predictive value.