A simple one-dimensional surface energy balance model, previously used to simulate the growth of urban heat islands on calm clear nights (Johnson et al., BLM, 1991), is modified to incorporate several additional controls on road surface temperature. In particular modules have been added to account for: horizon screening of solar and long-wave radiation exchange by buildings, trees and terrain; turbulent transport of sensible and latent heat; and cloud cover. The sensitivity of road surface temperature output to variations in several input conditions have been tested, including: view factors, deep subsurface temperature, substrate thermal properties, wind and cloud. Model predictions are validated against road surface temperature measurements at three sites in and near Vancouver, B.C. (urban centre, forested suburban, open agricultural) during winter. Results suggest that model predictions have an average error of about -0.5 Celsius degrees, and worst case errors of about ±1.0 degree.