Abstract: Abstract: Rational use of agricultural land is of significant importance to the food safety of our country and the development of social economy. With the introduction of new National Standards, classification of farmland has become an important part of land management. Accurate classification of farmland is the foundation of the management of both land quality and quantity. Classification of agricultural land is a very complex process, which requires a concurrent consideration of the natural quality, the utilized and economic level of land. Up till the present moment, the main method used to classify agricultural land was multi-factor comprehensive evaluation method, which was referenced on The Farmland Gradation and Classification（GB/T 28407-2012） issued by the Ministry of Land and Resources of China. The classification index of farmland in this method was calculated through a comprehensive assessment and progressive modification of natural, social and economic factors related to the land. The calculation of the index was based on light-temperature potential productivity and the standard farming system of the land. Although traditional methods did well in applying quantum mathematics to farmland grading, it had some drawbacks such as absolute quantity and rigid division, due to the complex process of classification of agricultural land. It was essential to explore a more scientific approach to classify agricultural land. The purpose of this paper was to apply fuzzy mathematics to classification of farmland, and explore the feasibility of combining fuzzy comprehensive evaluation with fuzzy clustering analysis, which was termed as the fuzzy comprehensive analysis method in the study. Take Anlu city as a study case, we employed the fuzzy comprehensive evaluation and fuzzy C-means clustering algorithm as the analysis methods. Besides, we used ArcGIS and Visual Studio2010 as the data processing platforms. The research process was as follows: Firstly, we obtained the degree of membership of grading units to each grade through compound operation of decision evaluation matrix and the weights of evaluation factors, based on the fuzzy comprehensive evaluation. And then, we used the membership matrix as a data source to grade agricultural land by fuzzy C-means clustering algorithm, since the method can make up for the information's loss caused by the principle of maximum membership in the process of classification. The results indicated that the fuzzy comprehensive analysis method can be used to classify agricultural land. Roughly 80% of grading cells of farmland were consistent with the result of traditional method. To further examine/test the accuracy of the method introduced in this study, we adopted the grain yield per unit of sown area in each village in Anlu city to verify the grading result generated by fuzzy comprehensive analysis method and the traditional method. Result showed that linear correlative coefficients were 0.87 and 0.82 respectively, which meant both methods had significant correlations. However, the correlation of fuzzy comprehensive analysis method was slightly higher than that of the traditional method. Thus, we can conclude that the application of the fuzzy comprehensive analysis method in classification of agricultural land can improve the classification of farmlands objectively and accurately, which has practical significance and reference value in quality evaluation of agricultural land.