Objectives: To evaluate benign and malignant thyroid nodules <= 1 cm according to ultrasonographic (US) features before surgery. Methods: Thyroid nodules 1-10 mm were evaluated in 600 patients. The US features in benign and malignant nodules were compared, and the odds ratios of the suspicious US features were determined with univariate and multivariate analyses. Diagnostic specifications such as sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and the area under the receiver operating characteristic curve value were calculated to evaluate the value of these US features. Results: Echogenicity, composition, height and width, margin, shape, capsule, calcification, and number were significantly different between benign and malignant nodules, whereas vascularity was not significantly different. Multivariate logistic regression analysis showed that solidity, taller than wide, irregularity, microcalcification, and multifocality were independent features in predicting malignant thyroid lesions. Irregularity had the highest sensitivity in all the independent features and solid composition was the most specific, although still at a low rate (only 69.1%). The sensitivity improved when two features were combined. Conclusions: Our study recommended that US features of solidity, taller than wide, irregular, microcalcification, and multifocality were useful sonographic criteria for differentiating malignant thyroid nodules <= 1 cm from benign ones. A combination of at least two of these characteristics can increase the diagnostic value for thyroid nodules.