BACKGROUND: Management of clinical T2N0M0 (cT2N0M0) esophageal cancer remains controversial. We reviewed our institutional experience over 21 years (1990-2011) to determine clinical staging accuracy, optimal treatment approaches, and factors predictive of survival in this patient population. METHODS: Patients with cT2N0M0 esophageal cancer determined by endoscopic ultrasound (EUS) were identified through a prospectively collected database. Demographics, perioperative data, and outcomes were examined. Cox regression model and Kaplan-Meier plots were used for statistical survival analysis. RESULTS: A total of 731 patients underwent esophagectomy, of whom 68 cT2N0M0 patients (9 %) were identified. Fifty-seven patients (84 %) had adenocarcinoma. Thirty-three patients (48.5 %) were treated with neoadjuvant chemoradiation followed by surgery, and 35 underwent surgical resection alone. All resections except one included a transthoracic approach with two-field lymph node dissection. Thirty-day operative mortality was 2.9 %. Only 3 patients (8.5 %) who underwent surgery alone had T2N0M0 disease identified by pathology: the disease of 15 (42.8 %) was found to be overstaged and 17 (48.5 %) understaged after surgery. Understaging was more common in poorly differentiated tumors (p = 0.03). Nine patients (27.2 %) had complete pathologic response after chemoradiotherapy. Absence of lymph node metastases (pN0) was significantly more frequent in the neoadjuvant group (29 of 33 vs. 21 of 35, p = 0.01). Median follow-up was 44.2 months. Overall 5-year survival was 50.8 %. On multivariate analysis, adenocarcinoma (p = 0.001) and pN0 after resection (p = 0.01) were significant predictors of survival. CONCLUSIONS: EUS was inaccurate in staging cT2N0M0 esophageal cancer in this study. Poorly differentiated tumors were more frequently understaged. Adenocarcinoma and absence of lymph node metastases (pN0) were independently predictive of long-term survival. pN0 status was significantly more common in patients undergoing neoadjuvant therapy, but long-term survival was not affected by neoadjuvant therapy. A strategy of neoadjuvant therapy followed by resection may be optimal in this group, especially in patients with disease likely to be understaged.