Regional climate scenarios are very important to the assessment of the impacts in the energy sector. Hydrological studies need climate scenarios to evaluate all possible ranges in the hydrological processes in future with respect to variations in greenhouse gas and sulphur emissions as well as demographic and economic developments. Existing techniques to project future weather could be successfully applied for one region or local place and would fail for another depending on the Global and Regional Climate Models behaviour, accuracy and presence of the observations, statistical analysis. Thus subsets of the complete suites of the climate scenarios for different regions could be determined by choosing the best methods to construct climate scenarios for these particular regions. A term Climate Scenarios Set ( CSS ) is introduced to reflect a necessity to simulate future climate using different technique : based on Global Climate Models ( GCMs ), Regional Climate Model ( RCM ) and simple climate models output, on temporal analogues related to variations in atmospheric circulation or spatial analogues, on downscaling etc. CSS is a product which could be applied directly to the impact models. CSS is developed handling the sources of uncertainty and including study of statistical significance of the results. SDSM ( Statistical Downscaling Model ) and LARS-WG ( stochastic weather generator ) are empirical statistical downscaling tools included into CCS development procedure. Comparative analysis is carried out to estimate the ability of these two tools to produce future climate for three time slices ( 1961-1990 as a baseline ) and present climate ( 1976-1990 ) with the 1961-1975 as a baseline. Merits and shortcomings of these statistical downscaling methods are indicated. SDSM and LARS-WG are recommended to be used for CSS development.