HVAC systems are different in terms of cost and performance. Their behavior in each single room or building is unique and depends on many factors, including the system type, the diffusers characteristics (the number, locations, and type of diffusers), the quality of dampers used, the thermostats' locations, the inside design temperature (thermostat setpoint temperature), and the control scheme associated with the HVAC system. Similarly, the building energy demand prediction is also unique, depending on the HVAC system design, the characteristics of building envelope and weather, people's behavior, and so on. This paper investigates the possibility of improving the heating energy demand calculation accuracy of a building equipped with an electric floor radiation system by the integration of a zonal model into a multi-room thermal model. Three thermostat setpoint strategies, room air temperature (RAT), mean radiant temperature (MRT), and operative temperature (OT) have been studied. Comparisons between the predictions of the thermal multi-room model and the integrated zonal model POMA/multi-room model developed have been performed to demonstrate the importance of considering the room temperature distribution in energy predictions. As well, detailed prediction of thermal comfort (PMV and PPD indices) allows the prediction of the relationship between thermostat locations, energy consumption, and thermal comfort.