With the global climate change and the aridification in northern China, more and more studies have focused on the landatmosphere interactions in arid/semiarid areas. The goal of this presentation is to evaluate land surface models BATS, LSM, and CoLM in semiarid areas using the Coordinated Enhanced Observing Period (CEOP) data, with an emphasis on comparison of the performance of the three models in semiarid regions. The results show that all of the three models can simulate the diurnal and seasonal cycle of the fluxes between land surface and atmosphere in this region, but the performance varies largely among them. For the simulations of the ground surface temperature and ground emitted longwave radiation, BATS is better than LSM and CoLM. As to sensible and latent heat flux, CoLM and LSM are better, and CoLM is the best. The performance of the three models varies with seasons, with summer better than winter. It is significant that all three models can not simulate the latent heat flux well in winter, which indicates that the current LSMs should pay more attention to improve the simulating performance in semiarid areas in winter.