Based on the periodicity of the pavement temperature change, the estimation model is built to calculate instantaneous temperature at any depth of the pavement structure, and also the layer average temperature from the surface to certain depth. By bringing in depth conduction time τz,, the model could narrow the discrepancy due to the temperature conductivity difference of pavement materials. According to one-year pavement temperature data collected by pavement temperature observation station from Guangzhou, Ningbo and Datong, the model is applied to confirm four coefficients by estimating the least residual standard deviation between calculated temperature and the measured temperature: t1 as the time length of temperature change in short time, t2 as the time length of baseline temperature change, k1 as attenuation coefficient of short-term temperature amplitude, and k2 the attenuation coefficient of baseline temperature amplitude. Fitting results show that the residual standard value between the calculated and measured temperature is about 1℃,no further than 1.52℃, and the average residual is less than 0.5 ℃, at any or certain depth of bituminous and concrete pavement in Guangzhou, Ningbo and Datong. Thus, the predicted formula is precise and commonly applicable.