Air infiltration has a significant impact on building energy performance and the indoor environment. Monitoring air infiltration continuously is of great importance to compare the airtightness of a building over time, and to detect building envelope degradation over time. An accurate estimate of air infiltration rate also informs envelope retrofit decisions to improve airtightness. However, air mobility and other environmental factors, such as wind or indooroutdoor temperature differences, often make the accurate measurement of air infiltration challenging. Further, conventional air infiltration testing approaches such as fan pressurization and tracer gas tests possess certain drawbacks limiting their applicability in commercial buildings. To address the limitations of air infiltration tests, this research proposes a low-cost inverse modelbased approach for estimating air infiltration rates by extracting the occupant-generated carbon dioxide (CO2) and humidity data from a building automation system (BAS). The laboratory tracer gas concentration decay tests were carried out to explore the effectiveness of replacing sulphur hexafluoride (SF6) with CO2 and the appropriateness of using low-cost BAS-grade sensors. Then the applicability of the proposed inverse model-based approach was verified by tracer gas concentration decay tests using both CO2 and water vapour. In this case, the historical CO2 and humidity data were used to validate the model to examine whether this approach can estimate infiltration rates from historical data. At last, return air CO2 concentration data from three air handling units were utilized to demonstrate this novel approach. Different regression models were developed to investigate the suitability of this ubiquitous sensor type to estimate building-level infiltration rates. The results indicated that the proposed method could conveniently lend itself to estimate air infiltration rates at a reasonable accuracy using existing sensor data.