Abstract: Abstract: Microclimate of swine house is a time-varying and nonlinear system affected by the interaction of internal production and external environment factors, and has an important effect on the pig's growth performance and health status. Among multiple environmental factors, indoor air temperature affects the heat balance of the pig body and plays an important role to maintain the constant body temperature. Humidity influences the evaporation of the pig body and therefore a body's thermoregulation. So, the indoor temperature and humidity are payed much attention by many researchers in the past decades in order to maintain a suitable indoor environment for pigs. Simulation of indoor microclimate is an effective way to provide a precision control strategy. Some researches on swine house environment control are conducted based on indoor air temperature and ventilation using traditional and automatic control method. However, in practice, it is difficult to realize an accurate microclimate control and prediction in a swine house due to some influences such as different regions that swine house located, seasons, raised number of pigs, and most simulations and models for swine house environment only focused on single environment factor such as temperature, humidity or airflow using computational fluid dynamics (CFD), the comprehensive interaction among the multiple factors and the energy consumptions are neglected. In this study, a thermal exchange model based on energy balance equations and a humidity variation model based on mass balance equations are developed for a closed swine house, the factors such as weather, heat dissipation, heat acquisition, building structure, building ventilation and pigs in the swine house are considered. Based on the one month's measured data of 1 minute resolution of room 2 on the south side and room 11 on the northwest corner in the swine house, some parameters of the simulation model were determined by using multivariate non-linear regression model (MNRM). Two days measured data are randomly selected to validate the model simulations from two rooms in hot summer. The energy consumptions of the two rooms are also analyzed and compared. Results show that the simulation and measurement values for both rooms agree well, the maximum temperatures error is 2.4 ℃, the maximum relative error is 9.2% for the two rooms, and the coefficient of determination is 0.836 9 for room 2 and 0.786 9 for room 11. The maximum relative humidity error is 13.34% and the maximum relative error is 49.66% for the two rooms, the coefficient of determination is 0.912 for room 2 and 0.899 7 for room 11. The power consumption of room 11 is 1.5 times that of room 2 because room 11 has extra heat produced by the pigs and a sidewall as part of the west wall of the building. The dynamic microclimate models based on the energy and mass balance equations can be used for simulations of basic environment control and energy requirement in closed swine houses.