CO2-based demand-controlled ventilation (DCV) has the potential to achieve energy saving while providing acceptable indoor air quality (IAQ). The spatial distribution of CO2 concentration could be non-uniform within an occupied space; however, limited information is available on how the sensor location affects the sensing performance of the breathing zone CO2 concentration. This study used experimentally validated Computational Fluid Dynamic (CFD) models to explore the optimal sensor position for the CO2-based DCV system considering varied ventilation system (displacement vs. mixing), air change rate, and occupant condition. The results show that the displacement ventilation causes vertical stratification of CO2, while mixing ventilation can achieve more uniform CO2 distribution. With mixing ventilation system, CO2 sensors located at the room return exhaust provide reasonable accuracies that meet the standard requirements. With displacement ventilation strategy, the errors of exhaust CO2 sensors exceed those requirements due to CO2 stratification and wall-mounted sensors placed at the breathing height yield smaller errors than the exhaust sensor. The results also indicate the accuracies of the CO2 sensors placed on the desk are unstable and vary significantly mainly due to impact of occupant thermal plume.