Vehicle traffic is one of the most significant emission sources of air pollutants in urban areas. While the influence of mobile source emissions is felt throughout an urban area, concentrations from mobile emissions can be highest near major roadways. At present, information regarding the spatial and temporal patterns and the share of pollution attributable to traffic-related air pollutants is limited, in part due to concentrations that fall sharply with distance from roadways, as well as the few monitoring sites available in cities. This study uses a newly developed dispersion model (RLINE) and a spatially and temporally resolved emissions inventory to predict hourly PM2.5 and NOx concentrations across Detroit (MI, USA) at very high spatial resolution. Results for annual averages and high pollution days show contrasting patterns, the need for spatially resolved analyses, and the limitations of surrogate metrics like proximity or distance to roads. Data requirements, computational and modeling issues are discussed. High resolution pollutant data enable the identification of pollutant "hotspots", "project-level" analyses of transportation options, development of exposure measures for epidemiology studies, delineation of vulnerable and susceptible populations, policy analyses examining risks and benefits of mitigation options, and the development of sustainability indicators integrating environmental, social, economic and health information.