Background: Leakage from esophageal anastomoses is higher than that for other gastrointestinal anastomoses. An intrathoracic anastomotic leak is a potentially catastrophic event. Methods: Patients with and without thoracic anastomotic leakage were compared for predisposing factors. Leak-related mortality was analyzed. Results: Of 475 patients, there were 17 leaks (3.5%). Predisposing technical factors occurred significantly more frequently in patients who leaked. Sixteen such events were identified as contributory in 11 patients. The hospital mortality for patients who leaked was significantly higher (35% versus 9%, P = 0.005). Inadequate drainage and persistent sepsis accounted for 4 of the 6 deaths. The need for inotropic support postoperatively correlated with leak-related mortality (66% versus 0%, P = 0.006), while leak size, time to diagnosis, or method of drainage did not. Conclusions: Thoracic anastomotic leaks are largely preventable. Leak-related mortality for the series was 1% and was most commonly related to inadequate drainage.