Robotic gastrectomy (RG) for gastric cancer (GC) has gradually gained nationwide prominence following 2011 guidelines from the Japan Society for Endoscopic Surgery (JSES), including the surgeons' criteria and the proctor system. In this retrospective study, we examined the short-term outcomes of the initial series of RGs performed by second-generation operating surgeons trained within our institute. Between January 2017 and April 2020, five surgeons each performed RG in 20 patients with clinical stage III or lower GC in accordance with the JSES guidelines. We evaluated both the rate of Clavien-Dindo grade II or higher morbidities and the console time required to reach the learning plateau via cumulative summation (CUSUM) analysis. We observed no mortality and 3% of morbidity following RG. Both the operative time (430 vs 387.5 min, P=0.019) and console time (380 vs . 331.5 min, P=0.009) were significantly shorter in the second 10 cases than in the initial 10 cases. We observed a remarkable trend in cases of distal gastrectomy (DG), in which the total operative time and console time were significantly shorter in the later cases. Our CUSUM analysis revealed that seven cases were required to achieve a learning plateau in RG when confined to DG. Non-expert RG surgeons meeting the operating surgeon's criteria from the JSES who had trained under an expert RG surgeon safely performed RG in an initial 20 cases.