Between December 2018 and May 2019, 90 patients evaluated with neck ultrasound (US) were included in this study. The thyroid nodules were categorized according to Kwak TI-RADS into 5 groups: TI-RADS 1, negative; TI-RADS 2, benign; TI-RADS 3, probably benign; TI-RADS 4a, low suspicion of malignancy; TI-RADS 4b, intermediate suspicion of malignancy; TI-RADS 4c, moderate concern but not classic for malignancy; TI-RADS 5, highly suggestive of malignancy. Pathological diagnosis was the reference standard in the evaluation of Kwak TI-RADS performance in malignant risk determination of thyroid nodules. Statistical analysis was done by measuring specificity, sensitivity, positive/negative likelihood ratio, positive/negative predictive values, accuracy, risk ratio, and ROC curve. The Fleiss and weighted kappa (κ) statistics evaluated the inter-observer agreement of Kwak classification of thyroid nodules and morphological features. Results: 90 patients with 225 thyroid nodules were included in our study. 191 (84.9%) were benign and 34 (15.1%) were malignant. A significant high malignant risk ratio was noted on the evaluation of the five main US features used in Kwak TI-RADS. The combination of TR4b, TR4c, and TR5 gave the highest area under the curve (0.989). The inter-observer agreement was perfect for Kwak TI-RADS categorization of thyroid nodules. Conclusion: Kwak TI-RADS has high diagnostic performance in the prediction of the malignant thyroid nodules.