Abstract: This study aims to overcome the shortage of the temporal variance and small coverage of TM images, which makes low accuracy for estimating the crop yield at province scale. The paper chose the maize yield in Shandong province in 2008 as the object of study. The data used in the paper mainly included six TM images with different phenophase of maize and the long time-series MODIS images with full coverage. The paper developed the time-series interpolating model based on the growth process of maize, which could interpolate the TM images with different phenophase into the dataset with the same milky maturity period of maize. Then through the in-situ measured samples of per unit area yield, the paper set up yield estimation models including ground-TM model and TM-MODIS model to obtain the full coverage yield information of maize at province scale. Results show that the method of yield estimation at province scale based on time-series interpolating model could make the most of the advantage of TM and MODIS data and avoid the regional disparity of NDVI derived from the temporal difference of TM images. The paper could reach high accuracy in the estimation of per unit area yield of maize in Shandong province. This will provide a new method to estimating crop yield at province scale.