By using energy balance method and considering the balance among solar short wave radiation, atmospheric and pavement surface long wave radiations, latent heat fluxes and sensible heat fluxes, a prediction model for AC road surface temperature is established. The pavement temperature and meteorological observation data of Ningbo city (from 2009 July 15 to 2009 August 13) have been used to validation of the model. The results show that a high accuracy was existed between the measured and calculated temperature in the clear days. when S (daily sunshine-ratio)>60%, the maximum absolute error of calculated daily maximum pavement temperature is less than 3 °C, and the multi-day mean absolute error is 1.5 °C. But in cloudy or rainy days, the model’s accuracy will be lower than the clear days.