e15028 Background: Ultrasound and ultrasound-guided fine needle aspiration (US-FNA) are a preferred method for judging benign and malignant thyroid nodules. For thyroid nodules that can’t be determined by FNA, molecular markers of thyroid cancer can be detected as assistant method. This study aim to verify the significance of molecular detection in diagnosis of benign and malignant thyroid nodules in China. Methods: We carried out a prospective study of 383 FNA samples of thyroid nodules, and performed targeted multi-gene NGS on most samples with FSZ-Thyroid NGS Panel V1, the result of which would be incorporated into the reference index for clinicians to decide whether a patient should undergo surgery. And postoperative pathology were followed up. The NGS targets were divided into three categories: high risk (BRAF V600E, TP53, PIK3CA, AKT1, CTNNB1, RET, KRAS with VAF≥30%, HRAS with VAF≥30%, NRAS with VAF≥30%, RET fusion, NTRK1 fusion, NTRK3 fusion, ALK fusion, BRAF fusion), low risk (BRAF non-V600E, EIF1AX, GNAS, TSHR, TERT, KRAS with VAF < 30%, HRAS with VAF < 30%, HRAS with VAF < 30%, PPARG fusion, THADA fusion), benign like (EZH1, SPOP, ZNF148 or no mutations). Results: Among the 383 FNA samples, 333 of them were sequenced with FSZ-Thyroid NGS Panel V1, and 211 thyroid nodules were resected. Finally, postoperative pathology showed that 180 nodules was malignant and 31 was benign. Among the 180 malignant tumors, BRAF V600E (71.1%), RET fusion (7.8%), and TERT mutation (2.8%) were the top three most mutated genes; and there were also significant differences in frequency of some mutations between malignant tumors in different BSRTC class, including more BRAF V600E (83.6%) in class VI malignant tumors, more gene fusions (26.1%) in class V malignant tumors which was significantly higher than that in class VI (7%). And, the frequency of RAS mutations (8.3%) in class III and IV malignant tumors was higher than that of class V and VI. Here a test was considered as positive if a genetic alteration was annotated with High-Risk, and negative if Low-Risk or Benign-Like, the sensitivity and specificity for detecting malignant tumors in BSRTC class III-IV were 66.7% and 100%, respectively, and the total sensitivity and specificity for BSRTC class I-VI malignant tumors were 85.6% and 100%. In addition, 21 of 383 patients underwent Re-FNA. Among these 21 patients, 11 patients detected high risk mutations and 90.9% (10/11) underwent upgrade of BSRTC class on Re-FNA. As a comparation, among the remaining 10 patients who detected low risk or benign like sites, only 30% (3/10) underwent upgrade of BSRTC class on Re-FNA which suggested that the detection of high risk sites might predict the upgrade of BSRTC class on Re-FNA. Conclusions: Molecular testing can distinguish between benign and malignant thyroid nodules and predict the upgrade of BSRTC class on Re-FNA which could provide the reference for treatment decision-making of thyroid nodules in China.