Introduction. The study of urbanization, agglomeration and growth of cities in Russia is very important. However, for spatially distributed data classical methods of regression analysis can give incorrect results and conclusions, so spatially inhomogeneous data necessary to apply special methods of analysis. Preliminary analysis. Group held cities in terms of population and analyzed the dynamics of population growth of cities in Russia, depending on their size and region (European part of Russia and Siberia and the Far East). The graphs of the dynamics of urban growth rates in population depending on the size of cities and regions. Revealed that the growth rate of urban population are not the same regions of Russia. A model proposed by Soo, with the inclusion of indicators of geographic market potential. Are graphs demonstrating the dependence of urban growth on the geographic market potential. Econometric analysis conducted logarithms urban growth in population in 2010 compared with 2002 is shown that spatial data, to account for all factors that influence the development of cities, it is necessary to use special methods of regression analysis. Method of investigation. For statistical analysis of data with spatial reference method used geographically weighted regression. Provides detailed mathematical description of the method of geographically weighted regression. Understand methods of constructing a matrix of weights, weight computation: administrative-territorial division, moving windows, fixed and adaptive kernels. When using the moving window Gaussian kernel addresses, bi-square and tri-cube. Discussion of results. A regression model of the market potential of Russian cities by geographically weighted regression. Scatterplot shows the logarithm of the predicted market potential based on constructed geographically weighted regression.