EUS has been shown to be the most accurate radiologic method of staging tumors of the gastrointestinal tract. EUS staging (Table 1) is critical in early esophageal cancer because of the potential impact on therapy. T0 and T1 cancers that do not penetrate deeply into the submucosa can be endoscopically treated with curative intent by using either endoscopic mucosal resection (EMR) or mucosal ablation with photodynamic therapy or argon plasma coagulation. T1 tumors are readily seen as focal thickenings of the mucosa (first and second layers) that may extend into the submucosal layer but not into the muscle layer. The accuracy rate for staging esophageal cancers is lowest for T1 and T2 lesions. Hence, highfrequency catheter miniprobes have been used to improve the staging of early cancers.1 Probes can raise the EUS T-staging accuracy for early squamous cell esophageal cancers from 71% to 92% (71% to 86% for T1 mucosal cancer and 78% to 94% for T1 submucosal cancer).1 However, there are little published data on the EUS staging accuracy in patients with early adenocarcinoma in Barrett’s esophagus. Earlier studies with firstand second-generation standard radial echoendoscopes suggested a low staging accuracy.2,3 However, one later study reported high sensitivity (100%), specificity (94%), and negative predictive value (100%) of preoperative EUS for detection of submucosal invasion.4 Preliminary data from studies with both standard and high-frequency EUS suggest that the presence of layer 3 disruption suggests submucosal invasion, but its absence does not exclude it.5 Indeed, in a large (n = 130) retrospective series from the Mayo Clinic,3 the overall accuracy of EUS T-staging was only 61% (43% for T1 and 70% for T0) because of both overstaging and understaging. The T-staging accuracy was not improved significantly in the 16 patients who were also preoperatively staged with the 20-MHz probe. The risk of lymph node metastasis approaches 40% in deep T1 submucosal esophageal cancers. Hence, EUS N-staging is also important in selecting patients for possible endoscopic curative therapy of early esophageal cancer. When EUS N-staging is compared with results of meticulous operative dissection of lymph nodes, the overall accuracy for staging lymph nodes in esophageal cancer is about 87% to 88%.6 But most patients reported in studies of EUS and esophageal cancer staging have advanced disease. There are little published data on the accuracy of EUS for determining the presence or absence of lymph node involvement in patients with early adenocarcinoma. Mediastinal lymph nodes can be commonly found in patients with Barrett’s esophagus. In one series of patients who had preoperative EUS, the prevalence of malignant, benign, and indeterminate lymph nodes in patients with high-grade dysplasia who had surgery was 0%, 25%, and 12%, respectively.7 Limited data suggest that the EUS accuracy for detecting malignant lymphadenopathy or excluding it is high. One small published study showed that preoperative EUS detected the single instance of lymph node metastasis in 1 of 17 patients (this patient had a T3 N1 cancer) but inaccurately predicted the presence of N1 disease in 4 patients.4 Hence, the specificity of EUS was 81% and the sensitivity was 100%. Our prospective data from Johns Hopkins support these findings.7 In 27 patients who had Barrett’s esophagus with highgrade dysplasia or early adenocarcinoma who had preoperative EUS, prevalence of malignant lymphCurrent affiliation: Division of Gastroenterology and Hepatology, Johns Hopkins University School of Medicine, Baltimore, Maryland.