Reliable modeling of emissions and dispersion of pollutants emitted in exhaust gases from motor vehicles is highly challenging due to the presence of multiple variables and inconsistencies in input data quality across different stages of the process. This article focuses on the vehicle fleet. It has been demonstrated that the structure of the vehicle fleet in a given area varies depending on the data collection methods and sources, which ultimately determines the dispersion results. This issue becomes particularly significant in urban environments, where the intensity of road traffic is high, and ventilation conditions are often poor, partly due to the formation of street canyons by dense urban development. The modeling was conducted at a selected intersection in Wrocław for various fleet structure scenarios. The results were compared with the results of pollutant concentrations from a nearby air quality monitoring station. The Copert emission model and emission factors from the European EMEP/Corinair database were used. In contrast, the GRAL model, a CFD model (suitable for urban dispersion modeling), was used to simulate pollution dispersion.