The role of interzonal airflows is especially pronounced in naturally ventilated buildings. In such buildings, reversed airflows in the ventilation stacks might occur as well. This affects the air exchange rate and contaminant distribution in buildings. A significant increase in carbon dioxide concentration is a characteristic phenomenon for poorly ventilated rooms. This paper demonstrates the application of metabolic carbon dioxide concentration measurements for interzonal airflow estimation in naturally ventilated buildings. The presented method is based on the continuous measurements of CO2 concentration at one point in each zone. These measurements are used to estimate airflow pattern in a multizone building by applying an inverse analysis. The developed methodology employs an iterative Levenberg-Marquardt procedure to maximise the nonlinear likelihood function. The validity of the method was verified against measurements carried out in a single naturally ventilated room. Further, the method was applied to calculate the airflow pattern in two apartments in Poland, containing 4 and 6 zones. The obtained results revealed very poor ventilation in both investigated apartments and reversed airflow in exhaust ducts. The amount of fresh air entering the rooms was insufficient to ensure good indoor air quality. The developed methodology can be effectively used as a diagnostic tool to identify the potential problems with ventilation systems.