Abstract: In order to classify the level of service(LOS) for urban rail transit station passage facilities, a classification method based on Gaussian mixture model is proposed. In which the per capita space of passages is taken as the classification standard, and a Gaussian mixture model of LOS for passage facilities is established based on the passenger flow, density, and speed in time. The model is used to classify the passage LOS. Finally, the actual data of passenger flow at Guangzhou metro Yangji Station is used to verify the model, and confirm the practicability of the model. This research provides a basis for the classification of LOS and passenger flow guidance of the station passage facilities.