In the city of Rio de Janeiro, for reasons of economic and administrative character, the number of measurement points of an air quality monitoring network is limited and, especially, the spatial arrangement of them may not have been studied carefully, being positioned in an unrepresentative place. Therefore, to support the concentrations measures and know their evolution, in this study was used models based on computational fluid dynamics. The area surrounding the Tijuca station was chosen as the study site and was investigated the dispersion of the pollutants; carbon monoxide, nitrogen dioxide and sulfur dioxide emitted by vehicular sources, in neutral atmospheric condition, more common in the winter months. It was used the commercial package Fluent version 17.1 and the RANS transport equations were closed with the standard K-epsilon turbulence model. Among the most important tasks can be mentioned: support a developed atmospheric boundary layer over the computational domain, develop the appropriate settings of boundary conditions, define the model constants, the modification of the law of the wall constants and perform the tests computational grid based on convergence criteria for an increase in the mesh density. The study allowed us to analyze the influence of the turbulent kinetic energy generated by the movement of vehicles on the streets on the behavior of the plume, and determine which is the turbulent Schmidt number that best fits the model. The results obtained in the simulations were compared each other and with available experimental data sensing in Tijuca monitoring station, observing good agreement data.