The incidence of esophagogastric junction (EGJ) adenocarcinoma has increased worldwide. As the EGJ is located at the boundary between the thoracic and abdominal cavities, the optimal surgical approach is a subject of debate and estimation of the esophageal invasion length (EIL) is an important factor in its selection. Data from our in-house database were extracted for consecutive patients with Siewert type I, II and III EGJ adenocarcinoma (EIL ≤ 4cm), who underwent transhiatal or transthoracic surgical resection between 2010 and 2016. The clinical records of these patients were reviewed and the accuracy of EIL estimation and its discrepancy with the pathological measurement were analyzed. A total of 82 patients were included in the final analysis. We established that EIL was underestimated in 49 of these patients (59.8%). The mean-distance discrepancy between the preoperative and pathological diagnosis of EIL in the underestimation group was 7.0mm. Multivariate analysis revealed that submucosal cancer spread was an independent risk factor for underestimation (P < 0.01). The mean length of submucosal cancer spread was longer for undifferentiated histologic type EGJ adenocarcinomas. (P < 0.01). The EIL was underestimated in approximately 60% of EGJ adenocarcinomas requiring surgical treatment. Thus, careful management is necessary, especially for EGJ adenocarcinoma of the undifferentiated histologic type.