Taking Jilin Province as an example, this paper explored an approach to risk zonation of agricultural natural disaster at kilometer level assisted by RS and GIS techniques. The index system characterizing the agricultural hazard, sensitivity and vulnerability was generated with RS and historical disaster data, and spatialized to kilometer-grid system. Then the risk zonation map was created by using analytical hierarchy process and linear weighting model. The paper also compared the zonation maps at spatial resolutions of 1km, 5km and 10km. The result indicated that: 1km-grid level was a preferable scale for the risk zonation of agricultural natural disaster in this region, in addition, Jilin Province was primarily of moderate and light risk, and extremely heavy and heavy risk areas distributed in the western and middle regions with area percentages of 9.56% and 18.15%, respectively.