One of the most obvious characters of freshwater ecosystems is their heterogeneity, namely, their disconnected structure and isolated functions, as comparing with terrstrial environments.However, this character is usually neglected by most traditional statistical methods, so using these tools can t describe these ecosystems correctly.In this paper, the authors recommend geostatistics to solve this problem-Based on regionalized variable theories, geostatistics is one kind of spatial statistical theory used to explore the correlativity and dependence between spatial variables.The first character of this method is its emphasis on the importance of spatial dependence of variables-In practical research, semivariance values of ecological factors or other indices can be calculated from the semivariance formulate according to the theory, and then, semivariogram can be drawn, distribution character of the research object (such as clumped or uniform pattern) can be found from the graph-Mathematical models simulation should be used in quantification of this character.Fractal dimension (D) obtained by double-logarithm semivariogram (D=(4-m)/2, where m (the slope of the graph) shows its heterogeneity character as well.The lower the value is, the higher heterogeneity the distribution of object.Kriging is another quantifying tool deduced from the theory.By kriging, the spatial concentration (density and so on) pattern of the variables can be known, values of unkown points also can be estimated, and this calculation value is more creditable because of its concern for regional property of the variables.In view of its accuracy, geostatistics would be widely used in freshwater ecology research such as density or spatial pattern of float grass, nitrogen, phosphorus and their ability of spatial occupation and so on.