A retrospective review of esophagectomy for esophageal carcinoma between 1982 and 1999 was performed. Two hundred twenty-two patients (mean age 61.7 years) underwent esophagectomy: 128 transhiatal, 74 Ivor Lewis, and 20 abdominal. Most tumors were adenocarcinoma (65%); the majority were in the lower third or cardia (78%). Excluding operative mortality the one-, 3-, and 5-year survival rates were 67, 39, and 31 per cent (median survival, 16.3 months) respectively. The hospital mortality rate was 6.8 per cent. Through univariate analysis race other than white, history of weight loss, poor or moderate differentiation (P = 0.05), full-thickness invasion (P = 0.02), positive lymph nodes (P < 0.01), Ivor Lewis esophagectomy (P = 0.02), intraoperative blood transfusion (P = 0.01), and tumor location in the upper or middle third in node-positive patients (P = 0.02) were associated with a poorer survival. Adjuvant therapy improved survival for patients with positive lymph nodes (P < 0.01). In multivariate analysis positive lymph nodes, tumor location, intraoperative blood transfusion, and adjuvant therapy were independent predictors of survival. To optimize survival esophagectomy for esophageal carcinoma should be performed without blood transfusion, and node-positive patients should receive multimodal therapy.