Cranial nerve system is the core part of human Nervous system. Different diseases on the pathway of the system correspond to different symptoms. According to the diagnosis of these symptoms the doctor can early diagnose out the related diseases on the nervous pathway. In order to master the adept skills, the neurologic medical staffs need to be trained on the diagnostic methods as well as learn abundant knowledge. As a traditional way of education, books and videos are used to help the students to comprehend the knowledge. However, compared with performing the examination on the real patient, this common way lacks vividness. Especially there are no active interactions. Recently more and more medical simulators are developed based on robotic technologies. These simulators bring innovative methods for medical education. Nevertheless, they just focus on the symptoms' reproduction, and do not have the abilities to show the relationships between sicknesses and symptoms. In this paper, we will propose a face robot (Fig. 1) for cranial nerves examination training to improve the training effectiveness. In this robot, we not only simulated various symptoms of optic tracking, eyeball movements, facial paralysis, but also made a cranial neurological model to simulate the different influences of the disorders on the nervous pathway. Finally, we performed a set of experiments to assess our proposed mechanism and control system. Three neurologists with over 30 years' experience were invited to show their opinions on the robot, and to discuss the results. All the results lead to the improvements for further researches.