Indoor temperature distributions and air flows lead to the variation of local thermal comfort from place to place. To have more precise predictions and better control of the thermal conditions in the working zone where the room occupant sits and works, a both detailed and fast model of the dynamic indoor temperature distributions is needed. Unfortunately, very few papers studied such models due to the complexity of fluid (air) flows. This paper discusses a zonal model which is derived from computational fluid dynamics (CFD) and the output of a CFD code. The model is validated with experimental results. In order to design better control systems, the zonal model is transformed into a state space representation form. One example is given on how the state space model can be used for temperature predictions and more precise temperature controls.