Land use spatial distribution has been well known to determine the pattern of urban traffic and the location of emission sources, thereby affecting urban air quality. This paper proposes a quantitative meth-odology to optimize the land use spatial structure and improve air quality by exploring the interrelationship between land use and traffic emission. The proposed model is able to optimize the air quality level as well as allocating land use following the bid-rent mechanism. As traffic emissions have been considered as a leading and determining contributor of air quality, the air quality is quantified by the pollutant emissions from both traffic and commercial/industrial activities in ur-ban areas. The land use model employs the bid-rent theory to describe the agents’ behaviors in the land market, taking consideration of envi-ronmental impacts. By using a combined genetic with frank-wolf algo-rithm, this study analyzes the relationship between urban land use spa-tial patterns and air pollutants emissions. In a case study of an empiri-cal urban area, the optimal land use pattern has been generated from the proposed model to achieve better air quality. In the optimal scenar-ios, the land use patterns have shown to be effective in improving air quality via the mixed and polycentric land development. Moreover, the results of residential spatial distribution indicate that the model can also effectively describe the bid-rent principles in the agents’ location choice considering environmental impacts.