Background The safety of robotic gastrectomy (RG) for gastric cancer in daily clinical settings and the process by which surgeons are introduced and taught RG remain unclear. This study aimed to evaluate the safety of RG in daily clinical practice and assess the learning process in surgeons introduced to RG. Methods Patients who underwent RG for gastric cancer at Kyoto University and 12 affiliated hospitals across Japan from January 2017 to October 2019 were included. Any morbidity with a Clavien-Dindo classification grade of II or higher was evaluated. Moreover, the influence of the surgeon's accumulated RG experience on surgical outcomes and surgeon-reported postoperative fatigue were assessed. Results A total of 336 patients were included in this study. No conversion to open or laparoscopic surgery and no inhospital mortality were observed. Overall, 50 (14.9%) patients developed morbidity. During the study period, 14 surgeons were introduced to robotic procedures. The initial five cases had surprisingly lower incidence of morbidity compared to the following cases (odds ratio 0.29), although their operative time was longer (+ 74.2 min) and surgeon's fatigue scores were higher (+ 18.4 out of 100 in visual analog scale). Conclusions RG was safely performed in actual clinical settings. Although the initial case series had longer operative time and promoted greater levels of surgeon fatigue compared to subsequent cases, our results suggested that RG had been introduced safely.