The areas of arable lands in most statistical surveys in China today are not the effective areas of arable lands. It contains the areas of many small land features. This paper presented a mathematic analysys method to compute the effective area of arable lands by deducting the area of small land features from the statistical data on the area of arable lands. The method integrated the relevant remote sensing (RS) and geographic information systems (GIS) technologies with a fuzzy GM (1, 1) model and a prediction model. Once the area of non-cultivatable small land features was calculated by the model, the effective area of arable lands was then computed by subtracting the calculated area of those land features from the total area of the arable lands. The method was empirically applied in Shunyi disdtrict, Beijing to test its applicability. The case study results indicated the method was effective with minimum labor, material and financial costs, and can be easily applied to other regions to calculate the effective area of the arable lands.