The aim of this study is to address the following issues: 1) revealing the typical behaviors and properties of pedestrian movement when going upstairs and downstairs; 2) constructing a pedestrian evacuation model to formulate the walking process of pedestrians in stair area; 3) verifying that the cell transmission model widely used in the two-dimensional walking space can also be applied to the three-dimensional staircase area. Firstly, an observation experiment is carried out to gain the pedestrian movement data in the process of going upstairs and downstairs. By collating the data, the relation between density and flow in the unidirectional process of going upstairs or going downstairs, and in the bi-directional process of going upstairs and downstairs, are drawn respectively. Then, by analyzing the fundamental diagrams, several characteristics of pedestrian movement in stair area are revealed. Based on these characteristics, an extended cell transmission model is proposed. In this model, a potential correction coefficient is introduced to change the route choice of pedestrians by using the influence of different directional pedestrians on the potential; a flow modification coefficient is introduced to describe the effect of physical parameters on the maximum flow at the boundary between two neighboring cells; and an offset coefficient is introduced to correct movement rules and strengthen the influence of preferential direction on pedestrian route choice. Further, simulations relied on the proposed model are conducted. By comparing the simulation results with the experimental data, the model is calibrated. Then the calibrated model is employed to formulate the pedestrian movement in stair area, and the sensitivity of the potential correction parameter is also discussed. The simulation results indicate that the proposed model can successfully reproduce the movement of pedestrians on stair. Moreover, the route-choice behaviors of pedestrians can be directed by varying the values of the potential correction coefficient, which can present important information about optimizing the evacuation process of pedestrians on stair, thereby reducing the risk of an accident, such as congesting and treading.