BACKGROUND: Esophagectomy for submucosal (T1b) esophageal adenocarcinoma (EAC) is performed in order to optimize patient outcomes given the risk of concurrent lymph node metastases (LNM). However, not seldom, comorbidity precludes these patients from surgery. Therefore, the aim of our study was to assess the course of follow-up after treatment in submucosal EAC patients undergoing surgery versus conservative therapy and to evaluate the incidence of metastatic disease. METHODS: Between 2001 and 2012, all patients undergoing diagnostic endoscopic resection for EAC in two centers were reviewed. Only patients with histopathologically proven submucosal tumor invasion were included. Submucosal EACs were divided into tumors that were removed radically (R0) and irradically (R1). Subsequently, in the R0 group, EACs were classified as either low risk (LR; submucosal invasion <500 nm, G1-G2, no LVI) or high risk (HR; deep submucosal invasion >500 nm, G3-G4 and/or LVI). Metastatic disease was defined as LNM in surgical resection specimen and/or evidence of malignant disease during follow-up (FU). RESULTS: Sixty-nine patients with a submucosal EAC were included [23 R1-resections and 46 R0-resection (14 R0-LR and 32 R0-HR)]. Twenty-six patients underwent surgical treatment (1 R0-LR, 12 R0-HR and 13 R1). None of the 14 R0-LR patients developed metastatic disease after a median FU of 60 months. In the R0-HR group and R1 group, metastatic disease was diagnosed in 16 and 30 % of patients, respectively. Surgical patients tended to have a better overall survival than non-surgical patients (p = 0.09). Tumor-related deaths, however, were 12 % in both groups. CONCLUSIONS: In LR submucosal EAC, the risk of metastatic disease appears to be very low. In deep submucosal EAC (either R0- or R1-resection), the rate of metastatic disease is lower than reported in earlier surgical series. Given the reasonable disease-free survival and high background mortality, conservative management of these patients seems to be a valid alternative for surgery in selected cases.