Abstract: The development of space technology and computer science pushes remote sensing (RS) and geographical information system (GIS) technology forward. Mathematical models (MM), as a link between RS, GIS, and their practical applications play an important role. It can be said that the progress of the application of MM is a key point for the development of RS and GIS technology from qualitative analysis to quantitative analysis, from static description to dynamic description and from single factor to multi-factors analysis. It is also a prerequisite to apply these new technologies in geoscience. Compared with traditional MM, those MM, applied in RS and GIS, have their own characteristics and superiority. Firstly, it can use the values of some variables in continuous space as parameters. This changed the situation that traditional MM can only use the average values of local points or a certain point to express some features, and overcome ,the defect that the accuracy is limited by the density of observation points, so that the simulation accuracy of MM is improved. Secondly, parameters of MM can be obtained in real-time or near real-time, while it has to take a long time to do ground studies for traditional MM to obtain those parameters, thereby it is made possible to do analysis in time and make dynamic simulation. Furthermore, it can make use of both the graphic and digital data stored in data base. Integrated analysis of various kinds of data, such as image data, terrain data, and thematic maps data can be conducted. Finally the calculated results of MM, obtained under the support of RS and GIS, can be expressed in two-dimensional or three-dimensional space, and we can observe current and even future possible situations intuitively. This paper is an overview of practical application examples of Chinese researchers in the area of the product estimation in agriculture, forestry, and animal husbandry, simulation of hydrologic process, dynamic estimation of desertification and soil erosion, land resources analysis and evaluation. Their research work is conducted by using MM, and with support by RS and GIS. After analysing the development process of computer recognition models of RS images, the author of this paper firmly believes that for computer recognition of RS images, transition from statistical models of recognition pixel by pixel, based on spectral data to comprehensive models, where spectral data, spatial relations of pixels, and many kinds of auxiliary data are used, will be realized widely. For analysis methods, MM will develop from accurate ones to those, where knowledge of judgement is used under the support of expert systems. This rep- resents the trend of development of computer recognition of remote sensing images. For agricultural production estimation model, the high accuracy multivariate regression winter wheat production estimation model by using vegetation index which is calculated by NOAA-AVHRR data and main meteorological factors has been set up to predict the winter wheat yield covering the area of 90 percent of all the wheat plantation area of China since 1984. In the estimation of forest timber volume, it is impossible to get satisfactory result by using vegetation index because the VI only represents the biomass of that year. The variables of forest features, including stand type, age group and conopy density, is made by using remotely sensed data. By the measurement of forest timber volume in some ground sample areas, non-linear mathematical model of above variables, interactive variables between them and forest timber volume is finally established. Remote sensing hydrological model is based on the statistical analysis of hydro-parameters by conventional hydrological model and data from remote sensing images and GIS. For example, the structure and parameters of the Sacramento model of continuous simulation of valley hydro-process have been adjusted. Valley area is divided into subareas. Each subarea is classified according to its undercover by remotely sensed data, and the parameters of the valley can be determined by conversion function. Remote sensing technology provides powerful means for the modelling analysis of land desertification. China has made success in the estimation of desertification process by establishing regional regression model, trend surface analysis model and Markovian model. The establishment of soil erosion model is mainly dependent on the airborne remotely sensed data. The main factors affecting soil erosion are determind by using theory of grey system. The mathematical expressions of small valley erosion model are conducted. The remote sensing mathematical models of land evaluation are also established in China. They mainly include: land resource evaluation, land sustaining ability, land planning, regional economic, social and ecological environment joint development planning etc. Multi-variate regression analysis, multi-target planning, expert system analysis and system dynamics are applied to set up the models.