Involvement of celiac nodes is defined as distant metastasis in the TNM classification for thoracic esophageal carcinoma. Some textbooks, however, describe dissection of these nodes as a standard technique. The present study was, therefore, undertaken to clarify which celiac nodes are regional for thoracic esophageal carcinoma and whether or not celiac node dissection would provide a survival benefit. Eight hundred and five patients who underwent R0 resection (no residual tumor) with systematic lymphadenectomy including the celiac axis area for thoracic esophageal carcinoma were retrospectively investigated. The frequency of metastasis and the therapeutic value of dissecting celiac nodes were compared to those associated with the left gastric artery area. The frequencies of left gastric and celiac nodal involvement were 15.4% and 9.6%, respectively, for thoracic esophageal carcinoma. As for tumor location, the incidences of metastasis around left gastric artery and celiac axis from the upper, middle and lower portion were 6.7% and 1.0%; 12.3% and 7.7%; and 25.7% and 17.4%, respectively. The 5-year survivals of patients with celiac but not left gastric metastasis were 36.3% and 41.8% for the middle and lower portions, respectively, while the corresponding values with left gastric involvement but no celiac metastasis were 24.1% and 27.9%. These differences were not significant. The frequency of celiac node involvement was not low. And, their dissection has equivalent therapeutic value to that of left gastric nodes. Revision of the TNM classification to account for celiac node involvement in thoracic esophageal carcinoma, especially of the middle and lower portions, is needed.