Cool roofs reflect sunlight and therefore can reduce cooling energy use in buildings. Further, since roofs typically cover about 20-25% of a city, widespread deployment of cool roofs could mitigate the urban heat island effect and partially counter urban temperature increases associated with global scale climate change. The magnitude of these potential benefits for a given city depends on the increase in albedo that can be achieved using reflective roofs. Assessing this increase requires knowledge of roof albedo at the city scale, which until now has been hindered, by a lack of reflectance data with sufficient spatial coverage, spatial resolution, and spectral detail. In this work we use multiband aerial imagery to derive the albedos of individual roofs in seven California cities: Los Angeles, Long Beach, San Diego, Bakersfield, Sacramento, San Francisco, and San Jose. The radiometrically calibrated, remotely sensed imagery has high spatial resolution (1 m) and four narrowband reflectances: blue, green, red, and near-infrared. First, we locate roof pixels within GIS building outlines. Next, we use laboratory measurements of the solar spectral reflectances of 190 roofing products to empirically relate broadband solar reflectance to reflectances in the four narrow bands; this empirical relationship well predicts solar reflectance, as indicated by a low root-mean-square of the residuals of 0.016. Albedos computed from remotely sensed reflectances are calibrated to ground measurements of roof albedo in each city. The error (accuracy) at 90% confidence interval of the calibrated albedos is found to vary by city, from 0.00-0.01 at low albedo and 0.06-0.14 at high albedo. (C) 2014 Elsevier Ltd. All rights reserved.