Natural ventilation is a common way of ventilation for urban residents, which is simple and energy-efficient. However, this ventilation not only introduces fresh air from outdoors into the indoor environment but also brings various pollutants from outdoors into the indoor environment, thus reducing indoor air quality and causing a series of human respiratory diseases in severe cases, such as asthma, pneumonia, bronchitis, etc. In a high-density urban environment, the proximity of rooms within a building can easily lead to cross-infection between occupants in the event of a public health emergency. Therefore, it is of great significance to quickly and accurately find the source of viruses or pollutants. The objective of this study is to accurately locate pollutant source which spread between units by wind effect. A model with three-storey building of wind-driven single-sided ventilation was built. Carbon dioxide (CO2) was used as a tracer of indoor pollutants. Computational Fluid Dynamics (CFD) is used to accurately simulate and predict airflow and concentration fields in and around the building. The results indicated that the location of the predicted pollution source is close to the position where the pollutant is released. The results of this paper can provide vital information for preventing the spread of contaminants in buildings.