PURPOSE: Since the approval of Siewert's classification during the 2nd International Gastric Cancer Congress in 1997, there has been no report on gastro-esophageal junction (GEJ) cancer using this new classification in Korea. The aim of this study was to assess the clinical usefulness of the new classification by applying it to the Eastern experience. MATERIALS AND METHODS: One hundred forty-six patients with an adenocarcinoma of the GEJ who underwent surgery from January 1990 to December 1998 were retrospectively classified according to the Siewert's classification. RESULTS: There were 2 patients (1.4%) with type I, 37 patients (25.3%) with type II, and 107 patients (73.3%) with type III. The proportion of the GEJ cancer among all gastric cancer patients who underwent operation was found to be 6.6%. The average age was 46 years in type I, 53 years in type II, and 55 years in type III. All types of GEJ cancers predominantly affected men. No significant differences in the pathologic variables including gross appearance, tumor size, and histologic findings were found between type II and type III. A curative resection (R0) was achieved in 118 cases (80.8%) without difference among tumor types. Type I tumors were treated with a gastrectomy with distal esophagectomy. Most of the type II tumors were treated with a total gastrectomy, and in some patients, a transhiatal partial esophagectomy was added. Type III tumors were treated with a total gastrectomy. Among resected cases with curative intent, microscopic tumor involvement of the proximal resection margin was noted in two patients (6.3%) with type II tumors. Lymph node metastasis was found in 2 patients (100%) with type I, 24 patients (64.9%) with type II, and 66 patients (61.7%) with type III. Lymph nodes along the lesser curvature were the most common site of metastasis, followed by pericardial nodes. The main lymphatic drainage directed the lymph nodes along the left gastric artery among the group 2 nodes. There was no difference in patterns of lymph node metastasis between type II and type III. Postoperative complications occurred in 29 patients, and operative mortality was 2.7%. Five-year survival rates were 0% for type I, 54.3% for type II, and 51.8% for type III. CONCLUSION: The distribution of subtypes of adenocarcinomas of GEJ was markedly different in this study compared with reported Western data. Therefore, Siewert's classification is useful for discussing GEJ cancer in an international setting as it clearly specifies the subtype of GEJ cancer. However, discrimination of type II and type III may have little clinical benefit since there were no significant differences in clinicopathologic characteristics or in the recommended types of surgery.