Objective The presence of extrathyroidal extension (ETE) is generally considered as an indication for total thyroidectomy for differentiated thyroid cancer. The accuracy of neck ultrasound for the diagnosis of ETE is controversial. The aim of this study was to assess the diagnostic accuracy of preoperative ultrasound evaluation of ETE. Methods The retrospective and observational study of consecutive patients operated between 2016 and 2019 for cytologically suspicious or indeterminate thyroid nodules were analyzed. US images obtained preoperatively were retrospectively reviewed to identify signs of minimal or gross ETE. Histology was considered as the golden standard for diagnosis of ETE. The sensitivity, specificity, positive (PPV) and negative predictive values (NPV), and accuracy of US were evaluated. Results A cohort of 305 patients (75% females, median age 48 years) with 378 nodules (median size 18 mm) was studied. Seventy-five percent of the nodules (n = 228) were malignant on histology and ETE was present in 106 cases (28%): 83 minimal ETE and 23 gross ETE. Suspicion of minimal ETE on preoperative ultrasound was found in 50 (13%) with a sensitivity of 30%, a specificity of 93%, a PPV of 62% and an NPV of 78%, with an accuracy of 76%. Gross ETE on ultrasound was found in 19 (5%) nodules with a sensitivity of 78%, a specificity of 99.7% a PPV of 94.7% an NPV of 98.6% and an accuracy of 98%. Conclusions Preoperative US is very specific and accurate in diagnosing gross ETE which impacts the extent of initial surgery for thyroid cancers.