Leakage, tumor recurrence, and stricture formation at the anastomosis are serious problems after esophagectomy for cancer of the esophagus or cardia. Because the prevalence of these postoperative complications may be affected by whether an anastomosis is made in the neck or in the chest, a comparison was made between anastomoses made at these two sites. During a period of some 7 years, we studied prospectively 411 patients who underwent resection for cancer of the esophagus or cardia and, after immediate reconstruction, had an anastomosis made in the neck or chest. The anastomotic leak rate for the neck anastomosis group was 4.3% and for the chest anastomosis group, 3.7% (p = not significant). The difference between leak rates of anastomoses fashioned by hand-sewn (5.0%) or stapled (3.0%) techniques was also not significant. The median upper resection margins in the neck and chest anastomosis groups were 4.5 cm and 3.5 cm, respectively. The corresponding rates of anastomotic tumor recurrence were 6.1% and 8.1% (p = not significant). The prevalence of benign stricture formation was significantly higher in the chest anastomosis group (19.2%) than in the neck anastomosis group (9.0%) (p = 0.002). This difference was a reflection of a significantly increased prevalence of stricture formation when an anastomosis was made by the stapler technique than with the hand-sewn method, and whereas most of the anastomoses in the neck were hand sewn (90%) the majority of those in the chest were stapled (80%). There were thus no statistically significant differences between the sites in terms of anastomotic leak and tumor recurrence rates, and the higher stricture rate in the chest anastomosis group was the result of more stapled anastomoses.