The adjustment of monetary policy and macroeconomic indicators (exchange rate of the national currency, refinancing rate, inflation rate) justifies the need to clarify the priorities of the tax policy and to consider possible ways of reforming the current Russian tax system. Hence an adequate assessment of the impact of possible changes in tax legislation on economic development is very important not only at the country level, but also at the level of its regions. However, most existing methodologies for assessing the impact of tax burden on the economy have considerable disadvantages. From the above, particularly important are issues related to the development of methods of economic tools to evaluate the impact of the tax burden on the gross regional product (GRP), which will provide more efficient construction of the budget and tax system of the country taking into account interests of the state and business in different regions. The purpose of the research is determination of critical values of the optimal tax burden in regions with different resource potential. Theoretical and methodological base of the research was the concept of the Laffer curve. The most important indicator of economic development of the area is its resource potential. The author of the article proposes the following elements of the resource potential of a region: labor, natural resources, innovation potentials, potential capital investment. It should be noted that basic labor and capital resources are present in every region, but natural and innovative resources are very unevenly distributed across the country. Taking into account this fact, the author develops a classification of the subjects of the Russian Federation in accordance with their resource potential. This classification involves division of regions into commodity and innovation regions. The group of commodity regions comprises four subgroups: mineral, land, forest and water regions. Using this classification, a model of the impact of the tax burden on GDP was developed, where the third factor (after labor and capital) for each classification group is its own indicator: natural (or innovation) resources. The model is based on a linear non-homogeneous production function, the quadratic functions of the tax burden are marginal productivities of factors. In theory, the proposed model is fitting, because the functions it presents behave satisfactorily at the extreme values of the tax burden, and these functions have all the properties inherent to a linear non-homogeneous production function. Based on this model, the critical values of the optimal tax burden in the subjects with varying resource potential are calculated. Thus, in the land regions the optimal level of the tax burden is approximately 13 %, in the innovation regions 17 %, and in the mineral regions 54 %. The results may be used by public authorities in determining the basic parameters of the regional tax systems.