The true value of endoscopic ultrasound (EUS) post-neoadjuvant chemotherapy for esophageal carcinoma is not established. Superior loco-regional detail may yield useful staging and prognostic information but information on its accuracy, as compared with computed tomography (CT), remains undefined and limited by small study size. We prospectively studied 109 patients with gastroesophageal cancer; 99 of whom were undergoing surgery. All had EUS and helical CT imaging before and after neoadjuvant chemotherapy and the results were compared with pathological staging of resected specimens. Tumor response was assessed by the reduction in maximal tumor depth at EUS and correlated with patient survival. There was no difference in T and N stage accuracies between EUS and CT following neoadjuvant chemotherapy. manova showed a reduction in maximal tumor depth by > 50% at EUS to be associated with longer survival (relative risk = 0.48, P < 0.05). EUS responders had a median survival of 38 months compared to 30 months for non-responders (P < 0.05). The identification of lymphadenopathy at radial EUS was not predictive of survival. This large series study demonstrates the staging accuracy of CT and non-biopsy EUS in the setting of neoadjuvant chemotherapy for gastroesophageal cancer to be equivalent and poor. An endosonography may contribute useful clinical information in respect of potential survival. It is questionable whether radial EUS should be included in protocols for restaging.