Since the Vreman subgrid-scale model was proposed, different types of flow fields have been used to research the applicability of this model. However, the applicability of the model to a flashover heat flow field simulation and the optimal model coefficient for room fires have not been reported. In this paper, the flashover room fire experiments of different opening sizes (ventilation factor) are conducted in an experimental room of size similar to ISO9705, and the room temperature at typical locations is measured. On the same numerical simulation platform, the room temperature field is simulated numerically using the large eddy simulation based on the Vreman and Smagorinsky subgrid-scale models. When the ventilation factors are 1.932 and 2.178, a Vreman subgrid coefficient of 0.02-0.04 is more applicable to the simulation of the temperature field. Overall, the simulation results of the maximum temperature based on the Vreman subgrid-scale model are slightly better than that of Smagorinsky subgrid-scale model. Moreover, the simulation result of the temperature change according to the time from the Vreman subgrid-scale model is similar to the experimental result. However, near the corner of the door in the room, the numerical simulation results underestimate the maximum flow field temperature in the room's upper part. When the Vreman subgrid-scale model simulates the temperature field in the room with a subgrid coefficient of 0.04 and a ventilation factor of 2.263, the simulation results are better than the results from the Smagorinsky subgrid-scale model with a subgrid coefficient of 0.2. However, when the ventilation factor is 1.397 and the opening is relatively small, the large eddy simulation result based on the Vreman subgrid-scale model with a subgrid coefficient of 0.04 is worse than the results based on the Smagorinsky subgrid-scale model with a subgrid coefficient of 0.2. (C) 2015 Elsevier Ltd. All rights reserved.