Most HVAC systems are designed to supply air based on assumed (usually maximal) rather than actual occupancy, therefore often resulting in over-ventilation. The concept and theories of demand-controlled ventilation (DCV), which are to find better ventilation strategies according to actual occupancy, have been developed for more than two decades and have been applied to many situations. However, a certain type of room (i.e. short-term occupied room) seems to have been neglected in the literature of DCV. The aim of this study is to work out an energy saving strategy for such types of rooms based on time-dependent concentration of CO2. This study investigated the time-dependent nature of the concentration of gaseous contaminants by both theoretical and computational fluid dynamics (CFD) approaches, and then applied this model to CO2 cases. Compared with existing models in the literature, the new theoretical model developed in this study has taken ventilation effectiveness into account to avoid the “instant perfect mixing” assumption thus better fitting actual situations. The theoretical predictions by this model fit CFD simulation results very well. Based on this model, an energy saving strategy for short-term occupied rooms was then proposed. Two practical examples in this study show that ventilation rates could be reduced to 50%~80% of the standard rates.