This study considers how the configuration of land cover affects the urban heat island in Phoenix, Arizona. We consider both the composition of land covers and the configuration, e.g. shape and size of patches, to determine the effects of landscape pattern on the urban heat island. Using methods from landscape ecology, we quantified spatial structure and then assessed the correlation between land surface temperature and spatial configuration of land covers. Based upon high-resolution land cover data, we examine three built land use categories in Phoenix: mesic residential, xeric residential, and industrial/commercial. The results of Pearson’s correlation analysis and multiple linear models reveal that composition and configuration both explain the variance in land surface temperature. Configuration variables were significantly correlated with temperatures in both mesic and industrial/commercial areas. Inclusion of configuration variables improved the fit of linear models compared to models that only included composition variables.