The efficacy of an extended radical lymph node dissection for carcinoma in the thoracic esophagus is controversial. Results of a multivariate analysis using clinical data from 127 cases collected from 1982 to 1988 are reported. Twenty-seven of these patients underwent an extended radical (cervico-thoraco-abdominal: 3 fields) lymph node dissection which was recently developed in Japan, while others underwent a standard (thoraco-abdominal: 2 fields) lymph node dissection. They all had a locally-curative resection of the tumor through a right thoracotomy. In this study, 13 factors commonly affecting prognosis were examined: sex, age, cancer location, tumor length, radiographic type, depth of invasion, lymph node metastasis, tumor differentiation, postoperative radiotherapy, chemotherapy, operative risk, postoperative complications, and 3-field or 2-field dissection. Based on the survival-rate curves using Kaplan-Meier's statistics, the 3-field dissection was superior to the 2-field dissection. Moreover, when other prognostic factors were adjusted using Cox's proportional hazards general linear model, the same result was obtained from survival-rate curves. From this analysis, it can be concluded that a 3-field dissection is a better approach for management of carcinoma in the thoracic esophagus.