Our research group is currently developing an endoscopic surgical robot for digestive organs. In the current study, we sought to train surgeons to manipulate the system we are developing for clinical applications. To this end, we are developing a training system with the same interface as the real system, so that surgeons in training can practice basic manipulations and surgical techniques using organ models. To learn the basic manipulations of the system, we emphasized training the surgeon to operate the robotic arms, as this is the biggest difference from the conventional surgical techniques. We set up several types of tasks for the trainee, so that a beginner trainee could get used to operating the robot arms of the system. We developed a surgical training method using a stomach model reconstructed from MRI data sets. In addition to basic surgical techniques such as grabbing, lifting and cutting open soft tissue with the robot arm, we enabled the training system to perform techniques necessary for the surgical system, such as delivering water to the surgical field in case of bleeding, and clipping of incision sites. We added a function to record the performance of the trainee, enabling the system to analyze changes of the surgical field and robot arms in four dimensions during training.