Background. Patient selection for surgery after neoadjuvant therapy for locally advanced non-small cell lung cancer depends on accurate restaging of mediastinal (N2) lymph nodes. Our objective is to assess the accuracy of endobronchial ultrasound (EBUS) for restaging N2 lymph nodes after neoadjuvant therapy. Methods. This is a retrospective review of patients with non-small cell lung cancer who underwent staging with repeat computed tomography and positron emission tomography and had restaging EBUS for sampling of N2 lymph nodes. Endobronchial ultrasound was performed for suspicious nodes in stations 2R, 2L, 4R, 4L, and 7. Selected patients who were N2-negative underwent thoracotomy with complete thoracic lymphadenectomy. Results. There were 32 patients with N2 disease who underwent preoperative chemotherapy or radiotherapy, or both, and subsequently had restaging EBUS. There were 3 patients who had recalcitrant N2 nodal disease detected by EBUS. There were 5 patients with pulmonary function or comorbidities that were prohibitive for surgery. Of the remaining 24 patients with negative EBUS, 3 underwent mediastinoscopy and 2 had recalcitrant N2 disease. The remaining 22 patients underwent thoracotomy. Recalcitrant N2 disease was noted in 1 patient at thoracotomy in the EBUS-assessable nodal stations. Thus EBUS was falsely negative in 3 patients. The sensitivity and negative predictive value of restaging EBUS were 50% and 88%, respectively. Conclusions. Restaging EBUS is relatively accurate at predicting the absence of metastatic disease in N2 mediastinal lymph node in patients who underwent neoadjuvant therapy for non-small cell lung cancer. © 2014 by The Society of Thoracic Surgeons.