Based on daily precipitation data of 2200 observation stations over China in 2009, precipitation at grid point is estimated by the method of moving window ordinary kriging and three technical experiments are designed in this article. The interpolation errors are compared between the overall search method and direction neighborhood method in the moving window ordinary kriging. The different effects of maximum influence radius and shielding on the interpolation error are also discussed. The sample point selection strategy of the direction neighborhood method is put forward to sort data points. The results show that compared with the overall search method for the traditional moving window ordinary kriging (MWK), direction neighborhood method has the notable advantages of more computational efficiency and high accuracy especially in the areas with dense sites. The results also show that good interpolation effect can be attained within the range from 4° to 5° meridional (zonal) arc length for most areas; and the shielding effect of precipitation is weak. Further more, better accuracy for precipitation estimation is presented when the ε value is equal to 0.1, but it is likely to get worse with the increase of ε value.