Since 1985, 229 cases of carcinoma of the esophagus have been considered for entry into a protocol with the use of preoperative chemotherapy and radiation therapy followed by surgical intervention as the primary element of treatment. One hundred sixty-five patients (93 with adenocarcinoma and 72 with squamous cell carcinoma) had esophagogastrectomy. The 5-year survival of the protocol patients who underwent resection was 25% for both groups--squamous cell carcinoma and adenocarcinoma. Of the protocol patients with squamous cell carcinoma who underwent resection, 40% had a sterilized specimen, whereas of those with adenocarcinoma, 20% had a sterilized specimen. If the patient had a sterilized specimen, the 5-year survival was approximately 60% for adenocarcinoma and 40% for squamous cell carcinoma. Those patients with adenocarcinoma and Barrett's esophagus had a 5-year survival of 55%. Of the patients who underwent only esophagectomy and esophagogastrectomy and had not been entered into the protocol, none lived beyond 3 years. The operative mortality rate for those who had esophagogastrectomy was 5%. Sixty-four patients completed the radiation therapy and chemotherapy but did not undergo surgical procedures because of progressive disease or refusal. Of those patients who completed chemotherapy and radiation therapy without surgical intervention, 5-year survival was 18% in patients with squamous cell carcinoma, whereas no patients with adenocarcinoma survived beyond 3 years. The finding of a sterilized specimen after esophagectomy is a favorable prognostic factor in patients with adenocarcinoma or squamous cell carcinoma. The finding that patients with Barrett's esophagus and adenocarcinoma have an improved chance for survival is perhaps related to an earlier diagnosis. It is clear that some patients with squamous cell carcinoma who did not undergo surgical procedures did have a sterilized specimen, because the survival in this group approached 20% at 5 years.