Background: Proximal gastrectomy is frequently selected for early gastric cancer located at the upper third of the stomach. Annual postoperative endoscopy is considered to be useful for the early detection of metachronous remnant gastric cancer (MRGC). Despite that, we sometimes have also encountered cases that are not suitable for endoscopic resection (ER). The aim of this study is to identify the risk factors for the patients who developed MRGC after proximal gastrectomy, received annual endoscopy, but are not eligible for ER (non-ER). Methods: We retrospectively analyzed 191 patients who had received annual endoscopic surveillance after undergoing PG for cT1 gastric cancer at NCCH between 2006-2015. MRGCs were categorized into two groups: ER group and non-ER group. The remnant stomach was defined as three locations: pseudo-fornix (PF), corpus and antrum. Results: The median observation period was 73 (range 12-168) months. MRGC was observed in 29 cases with 32 lesions. ER was eligible for 20 lesions (ER group, 62.5%), but 12 were not (non-ER group, 37.5%). We compared the characteristics between of ER group and non-ER group in terms of sex, age (>65 years), interval from the initial surgery (>60months), the location (PF vs non-PF), and cross-sectional circumference (posterior wall vs. others). In both univariate and multivariate logistic regression analysis, only the location at PF was identified as a risk factor for non-ER group (OR 25.0, 95% CI: 1.8–339, p = 0.015). Among 12 lesions of the non-ER group, the location of PF was found in 6 (50%) lesions. As compared with MRGC at the non-PF, that at the PF (n = 6) was characterized by younger patients (57.5 (range 37-69) years vs. 69 (56-77) years), larger size (36 (7-57) mm vs. 9 (9-24) mm), deeper depth (MP or deeper; 5/6 vs. 0/6), more frequent undifferentiated-type histology (4/6 vs. 0/6) and frequent nodal metastasis (3/6 vs. 0/6). The interval to detect MRGC appears to be longer in the PF (78 (31-96) months than that of the non-PF (60 (31-96) months), but the difference was not statistically significant (p = 0.86). Then, we examined the visibility and the mucosal normality of the area developing MRGC in the annual follow-up endoscopy one year before the detection of MRGC (n =32). In 25 lesions at the non-PF, the visibility and the mucosal normality was secured in 21 (84%). while 5 PF lesions were difficult to observe (71.4%, p = 0.01) due to food residues in 4 (57%, p = 0.047) and insufficient expansion of the gastric mucosa in 4 (57%, p = 0.001). Conclusions: Detection of the MRGC at the PF at an early stage is crucial and challenging. With annual follow-up endoscopy, complete clearance and adequately extending the gastric mucosa could lead to the early detection of MRGC at the PF.