Abstract: Abstract: The quantity and quality of cultivated land plays an important role in the national food security and the long-term stability of the community. The purpose of this study was to explore a method of location sampling with optical number for monitoring the quality of cultivated land at county level. Songyang County, Zhejiang Province was selected as our study area to investigate the proposed method. We adopted a combination of theoretical modeling and empirical analysis together in this study. The absolute deviation and the variance of national natural index of cultivated land were used to calculate the theoretical sample size of the monitoring points of cultivated land within the permitted statistical error range. We preliminarily determined the number of monitoring reference samples according to the technical specification for soil environmental monitoring. There were 47 monitoring sample points simulated in Songyang County. Then, we applied the variogram model of geostatistics to set up the monitoring sample points, which provided a way for controlling monitoring costs and improving monitoring accuracy based on variability analysis. The input parameter of semi-variable function was national natural index. All the process was explored in the software GS+. The semivariance of agricultural land natural index was analyzed at separation distance. The variation of the radius of cultivated land in Songyang was 3.1 km, which was the exactly radius of the grid in the proposed method. There were 89 monitoring points of land quality grading arranged by the use of variogram results. Next, sample points were corrected and optimized by the use of grading combination and spatial attributes. The core idea of the procedure was how to correct the monitoring sample points to reflect the land quality level successfully. In order to improve the representativeness of monitoring points and the integrity of arable land quality monitoring information, we analyzed the distribution characteristics of grading factors and the potential change areas of cultivated land quality. In this study, there were 14 grade combinations of national agricultural land natural level, national land use level and national agricultural land economical level. After spatial overlay analysis of quality grading factors, potential change areas and grade of cultivated land using the geostatistics method of the GIS (geographic information system), the number of reasonable layout of the monitoring points was reduced to 40. Finally, we compared the grade monitoring point arrangement and the distribution map of grading results of land quality respectively between the proposed and the traditional method. The overall accuracy of national natural grade, national usage grade and national economic grade was improved by 12.75%, 5.3% and 10.55%, respectively. Kappa coefficient was also improved by 0.19, 0.10 and 0.17, respectively. Additionally, the area ratio of grading factors was almost the same as the number ratio of monitoring points of cultivated land. We also analyzed the distribution of monitoring points within potential change areas of cultivated land quality. The area ratio of potential change areas was almost the same as the number ratio of monitoring points in potential change areas. Results indicated that the proposed method was effective for locating sampling for monitoring the quality of cultivated land and significantly improved the accuracy of the cultivated land grading. All the monitoring points were successful in representing the integrity of cultivated land quality. The explored new method is feasible, scientific and standard, which can be widely used in the management of monitoring the quality of cultivated land at county level. Further research is needed in other scales, such as city, province and country. The proposed method should be further tested in other regions containing more complex conditions in geology and geomorphology.