Objective To monitor surgical quality and analyze learning curve of minimally invasive totally thoracoscopic cardiac surgery. Methods We retrospectively analyzed the clinical data of 150 consecutive patients who underwent minimally invasive totally thoracoscopic cardiac surgery in the Guangdong General Hospital between January 2013 and December 2015. There were 60 males and 90 females at age of 43.1 years. There were 60 patients with atrial or ventricular septal defect repair, 12 patients with cardiac tumor resection, 53 patients with mitral valve replacement and 25 patients with mitral valve repair. According to the surgical sequence, all the patients were divided into 3 groups including a group A, group B, and group C with 50 patients in each group (every 10 patients as a sequence, every 5 sequence as a group). Surgical outcomes were compared among the 3 groups, and surgical quality was analyzed with descriptive statistics. Results Surgical failure rate was 6.7% (10/150). There was no in-hospital mortality. Aortic cross-clamp time, cardiopulmonary bypass time and duration of mechanical ventilation, duration of ICU stay, duration of hospital stays of the group C were significantly shorter than those of the group A and group B. Analysis showed a significant learning curve effect in totally thoracoscopic cardiac surgery. When surgical cases reached about 100 cases, cardiopulmonary bypass and aortic cross-clamp time was shorter than the average value stably. Conclusion Totally thoracoscopic cardiac surgery is safe and reliable. For the beginners, it needs about 100 patients of surgery to master the totally thoracoscopic cardiac surgery.