BACKGROUND The role of endoscopic ultrasound in evaluating the response of esophageal cancer to neoadjuvant chemotherapy is controversial. OBJECTIVE To evaluate the accuracy of EUS in restaging patients who underwent NAC. METHODS The disease stage of patients with esophageal cancer was established by means of the TNM classification system. The initial staging was determined by chest and abdominal computed tomography and EUS. Patients who needed NAC underwent a preoperative regimen consisting of cisplatin and fluouracil. Upon completion of the chemotherapy, patients were restaged and then underwent esophagectomy. The results of the EUS staging were compared with the results of the surgical pathology staging. This comparison was done in two groups of patients: the study group (all patients who received NAC) and the control group (all patients who underwent primary esophagectomy without NAC). RESULTS NAC was conducted in 20 patients with initial stage IIB and III carcinoma of the esophagus (study group). Post-chemotherapy EUS accurately predicted the surgical pathology stage in 6 patients (30%). Pathological down-staging was noted in 8 patients (40%). However, the EUS was able to observe it in only 2 patients (25%). The accuracy of EUS in determining the T status alone was 80%. The accuracy for N status alone was 35%. In 65% of examinations the EUS either overestimated (35%) or underestimated (30%) the N status. Thirteen patients with initial stage I-IIA underwent primary esophagectomy after the initial staging (control group). EUS accurately predicted the surgical pathology disease stage in 11 patients (85%). CONCLUSIONS EUS is an accurate modality for initial staging of esophageal carcinoma. However, it is not a reliable tool for restaging esophageal cancer after NAC and it cannot predict response to chemotherapy.