As shown in "Balance of industrial reserves and underground water resources in Poland", there have been significant losses in hard coal resources throughout 1989-2005 i.e. since the beginning of our political and economic transformation. Annual loss amounts were different in each kind, and the causes there of were not always clear. Statistical prognoses about sufficiency of resources and reserves was based on analysis of linear trends determined for geological resources and industrial reserves in Upper Silesian Coal Basin (USCB). Such prognosis is necessary in order to ensure power security in the country and to choose a proper strategy Prognoses were made in two variants. Variant A was based on data referring to 1989-2005 i.e. the entire period of economic and political transformation in the country. Variant B referred to 2001-2005 i.e. the period of significant improvement to economic situation in respect of coal. First stage of the analysis started with linear regression function. Then we verified the accuracy of correspondence between linear model and empirical data. This was based on the following parameters: standard deviation of the remainder component, remainder variation coefficient, convergence coefficient, linear determination coefficient. Next in the paper there tried to settle reasons for total decrease of the resources - deleting resources/reserves in a register and reclassification of such resources/reserves in 1989-2003 years. It was provided basing on characteristics of the resources. The study intended to present coal value and structure base of designated coal mines. Furthermore, there have also defined reasons for reclassifications and decreases of the coal resources, and verified the reliability of such resources/reserves. The following coal mines and beds were subject to the examinations: Bielszowice Hard Coal Mine - the seam No. 502, Pokój Hard Coal Mine - the seam No. 510/1, Halemba Hard CoalMine - the seam No. 340 and Polska-Wirek Hard CoalMine - the seam 510/1. Analysis of the trends indicates that the changes will take a non-linear route Therefore it is reasonable to expect that a prognosis based on linear function will not be accurate. Is it reasonable then to look for another, non-linear regression function? No, it is not because thereare no scientific premises (e.g. of physical or other nature) to suggest how such curve should go. Therefore the form of the trend should be assumed on realistic basis, in accordance with common sense. What can be done in this situation is to accept a less accurate prognosis on the basis of a linear decreasing trend. There are downward trends in losses of resources/reserves. Seventeen-year prognosis (variant A) shows faster loss of resources/reserves (higher depletion rate) than in variant B. This indicates that downward trends have been clearly inhibited during the last five years. The analysis of yearly increases or decreases of the resources in 1989-2003 years presented the total decrease of the resources/reserves. The most important reason to undertake a decision regarding reclassification of plots (mainly for out-of-balance resources of group "b") were natural reasons - complicated tectonic disturbances. But the most essential reason of deleting plots was the necessity of separate rendering accessible and complicated tectonics, i.e. subjective reasons.