In metropolitan regions, urban form is usually associated with pollutant levels. However, empirical analyses have relied on small samples and demographic variables of urban form. In this paper, we study the relationships between urban form indicators derived from land cover data and emissions for all of the conterminous United States. We use a spatial error model to identify to test if the urban form indicators, and in particular mixing of urban and forest land cover has effect on emissions. We find that fragmentary patterns are associated with higher emissions but these emissions are mitigated when there is a mixing of forest and urban land cover. This points to a nuanced understanding of the implications of sprawling city development on air quality.