Abstract: By collecting the subjective and objective data of the old metro station fire evacuation process and analyzing various influencing factors of the evacuation process, spatial optimization suggestions for the target test station are put forward. By comprehensively adopting technical means such as virtual reality, eye tracking and questionnaire survey, taking the fire evacuation process of Tianjin Metro Xi’nanjiao Station as the research object, the station modeling and fire scene rendering are carried out, and three evacuation tests of different difficulty levels are conducted on test personnel. The test personnel behavior data, eye tracking data, post test questionnaire survey data are recorded, and through comprehensive statistical analysis of these data, the influence of factors including individual behavior, basic information, spatial factors on evacuation is explored. Results show that gender of the test personnel and their evacuation knowledge apprehension level will affect the evacuation results. Homogeneous space will affect position judgment, and large volume structure blocks eyesight; interchange space is the problematic node during evacuation. Complex spatial experience and excessive guide signs make it difficult for test personnel to quickly identify the correct escape path. In the weak link of evacuation, spatial guidance, signage guidance and lighting guidance have a greater impact on evacuation, while factors such as other equipment have less.