Objective     To explore the clinical efficacy and learning curve of robot-assisted thymectomy via subxiphoid approach. Methods    The clinical data of patients with robot-assisted thymectomy surgery via subxiphoid approach performed by the same surgical team in the Department of Thoracic Surgery of Shanghai Pulmonary Hospital from February 2021 to August 2022 were retrospectively analyzed. The cumulative sum (CUSUM) analysis and best fit curve were used to analyze the learning curve of this surgery. The general information and perioperative indicators of patients at different learning stages were compared to explore the impact of different learning stages on clinical efficacy of patients. Results    A total of 67 patients were enrolled, including 31 males and 36 females, aged 57.10 (54.60, 59.60) years. The operation time was 117.00 (87.00, 150.00) min. The best fitting equation of CUSUM learning curve was y=0.021 2x3–3.192 5x2 +120.17x–84.444 (x was the number of surgical cases), which had a high R2 value of 0.977 8, and the fitting curve reached the top at the 25th case. Based on this, the learning curve was divided into a learning period and a proficiency period. The operation time and intraoperative blood loss in the proficiency stage were significantly shorter or less than those in the learning stage (P<0.001), and there was no statistical difference in thoracic drainage time and volume between the two stages (P>0.05). Conclusion    The learning process of robot-assisted thymectomy via subxiphoid approach is safe, and this technique can be skillfully mastered after 25 cases.