This paper aims to propose an evaluation method based on eye movement for motion sickness caused by virtual reality movement mechanism. When subjects experienced in the experiment, eye-movement data were collected in real time under the usability testing method called eye-tracking retrospective think-aloud. Subjects need to complete evaluation scales twice before and after the experiment and then 11 parameters were collected. To describe more clearly, the entropy weight method was used to improve the exploratory factor analysis to comprehensive evaluate. This method firstly extracted factors to simplify the number of variables and then corrected the weight of component factors by entropy evaluation. This new component variables had highly practical significance, and we obtained the virtual reality motion sickness evaluation model. The paired t-test results were greater than 0.05, which is effective. Based on the eye-movement tracking technology, the subjective motion can be described by objective eye movement data. Combined with VR, eye-tracking technology can collect data as feedback information to improve virtual reality motion mechanism. Also, it can further alleviate virtual motion sickness symptoms of subjects.