PURPOSE: To retrospectively evaluate the diagnostic performance of power Doppler ultrasonography (US) in depicting vascularity and to determine whether the combination of vascularity and suspicious gray-scale US features is more useful in predicting thyroid malignancy than are gray-scale features alone. MATERIALS AND METHODS: This was an institutional review board-approved retrospective study, with waiver of informed consent. A total of 1083 thyroid nodules in 1024 patients (median age, 51 years; range, 16-83 years), consisting of 886 women (median age, 50 years; range, 16-83 years) and 138 men (median age, 53 years; range, 19-74 years), were included. Nodules with marked hypoechogenicity, noncircumscribed margins, microcalcifications, and taller-than-wide shape were classified as suspicious according to gray-scale US criteria. Vascularity was classified as none, peripheral, and intranodular. The diagnostic performance of gray-scale and power Doppler US features was compared and classified as follows: criterion 1, any single suspicious gray-scale US feature; criterion 2, addition of any vascularity as one of the suspicious features to criterion 1; criterion 3, addition of peripheral vascularity to criterion 1; criterion 4, addition of intranodular vascularity to criterion 1; criterion 5, addition of no vascularity to criterion 1; criterion 6, American Association of Clinical Endocrinologists and Associazione Medici Endocrinology guidelines--all hypoechoic nodules with at least one of the following US features: irregular margins, intranodular vascular spots, taller-than-wide shape, or microcalcifications. RESULTS: Of 1083 nodules, 814 were benign and 269 were malignant. Intranodular vascularity was frequently seen in benign nodules and no vascularity was more frequent in malignant nodules (P < .0001, respectively). The area under the receiver operating characteristic curve (A(z)) of criterion 1 was superior (A(z) = 0.851) to that of criteria 2 (A(z) = 0.634), 3 (A(z) = 0.752), 4 (A(z) = 0.733), 5 (A(z) = 0.718), and 6 (A(z) = 0.806) (P < .0001). CONCLUSION: Vascularity itself or a combination of vascularity and gray-scale US features was not as useful as the use of suspicious gray-scale US features alone for predicting thyroid malignancy.