Abstract Although anastomotic leak is a common postoperative complication following an esophagectomy, it is not well known whether anatomical factors increase the risk for anastomotic leak after the procedure. The purpose of this study was to clarify whether a narrow thoracic inlet is an independent predictor of cervical anastomotic leak after retrosternal reconstruction following esophagectomy. Methods A total of 212 patients who underwent esophagectomy with gastric conduit retrosternal reconstruction between January 2013 and March 2019 were included in this study. Computed tomography was used to measure the thickness of the sternum (TS), the thickness of the clavicle (TC), the interclavicular distance, the sternum-trachea distance (STD), the sternum-vertebral body distance (SVD), and the sternum-trachea distance/sternum-vertebral body distance ratio (STD/SVD ratio). The correlation between various factors was analyzed using Spearman’s correlation coefficient. Tree-based analysis was performed to define cutoff values. Multivariate logistic regression was used to analyze the association between various predictors and anastomotic leak. Results Anastomotic leak occurred in 26 patients (12.26%). Tree-based analysis identified an optimal TS cutoff value of 20.84 mm, a TC cutoff value of 23.63, and a STD/SVD ratio cut off value of 0.2138 to predict anastomotic leak. There were significant associations between the STD, STD/SVD ratio and thoracic inlet area (STD × ICD). According to multivariate analysis, STD/SVD ratio, TS, TC, and diabetes mellitus were significantly associated with increased incidence of anastomotic leak. Conclusion STD/SVD ratio, TS, TC, and diabetes mellitus were associated with higher rates of cervical anastomotic leak after retrosternal gastric conduit reconstruction following esophagectomy. In patients with a small thoracic inlet, posterior mediastinal reconstruction and intrathoracic anastomosis should be considered.