Abstract: After the 9·11 incident, the research on fire-induced collapse of steel structures has been receiving a widespread attention from scholars all around the world. The physical and mechanical properties of steel material can be significantly reduced at elevated temperatures, thusly steel structures are prone to collapse under fire. This paper is based on the collapse experiment of planar steel frames with one column heated. The explicit dynamic analysis module of ABAQUS is used for simulation and the experiment is simulated by a beam element. The influence of high-temperature material model of steel on the simulation of collapse of steel frame structures under fire is studied. Some suggestions about how to choose the suitable high-temperature constitutive model and thermal expansion coefficient model for the simulation of collapse of a steel frame structure under fire are proposed. Three kinds of high-temperature constitutive models considering the effect of strain rate are enumerated. By comparing the simulation results with the experimental results, the effectiveness of each model in the simulation of collapse of steel frame structures under fire is analyzed. Ideal high-temperature constitutive model will lead to an inaccurate simulation of a critical temperature and failure mode, and there are some limitations in its application; EC3 high-temperature constitutive model can accurately simulate the quasi-static failure modes, but cannot simulate the dynamic failure mode correctly. Therefore, it is suggested to adopt EC3 or the improved EC3 high-temperature constitutive model for the simulation of a quasi-static failure mode. For the simulation of a dynamic failure mode, the improved EC3 high-temperature constitutive model is recommended and Dynamic Increase Factor (DIF) is adopted to consider the effect of a strain rate. In addition, by comparing four kinds of models of thermal expansion coefficients, the effect of phase transformation of steel material on the simulation of collapse of steel frame structures under fire is further studied. Although EC3 has provided a detailed thermal expansion coefficient model, the model proposed by GB 51249 is recommended as it more closely fits test results.