Abstract: The relationship between road surface temperature and weather conditions is a key problem for road surface temperature prediction which can prevent and reduce traffic accidents effectively. Many countries such as Canada, Denmark, and Germany have established road weather information systems and road surface temperature numerical forecast models while China is somew hat behind. From July 2006 to June 2007, observations are carried out at Meicun and Xianrenshan stations along the Huning Expressway when road surface temperature, airtem perature, humidity, wind direction, wind speed, and rainfall are recorded minutely. These data are analyzed to study the daily characteristics of expressway surface temperature for different seasons and weather conditions. On the basis of that, statistical models of minimal and maximal road surface temperature are established by stepwise regression for Meicun and Xianrenshan. It shows that daily variation characteristics of road surface temperature and air temperature are obviously different. The road surface temperature increases rapidly after sunrise and goes beyond air temperature when it is sunny or cloudy. While the discrepancy between road surface temperature and air temperature is smaller in raining days or at noon.In July, the discrepancy is extremely obvious. The simulated minimal road surface temperature is close to observations and errors are within ±2 ℃, so the model is applicable in practice. The errors for maximal road surface models are relatively large because of data limitation and not considering solar radiation impacts.The models established using statistical methods are local and need modifying when used in other regions. If more data are obtained, further research will be carried out to disting uish different seasons or weather conditions and the impact of solar radiation should be considered.