The focus of this paper is to explain a proper way of using climate model simulations in climate change impact studies. Special attention is addressed to reducing model bias, a systematic model error present in all climate integrations. Studies are done using climate projection results obtained by the Coupled Regional Climate Model EBU-POM, for three periods: 1961-1990 (experiment 20c3m), 2001-2030 (A1B scenario) and 2071-2100 (A2 scenario) for the Balkan area with special analysis of vineyard regions in Serbia. A Statistical Bias Correction method, based on daily values, is applied to the model results in order to minimize bias. Corrected values for 2 m air temperature and precipitation are used to calculate several climate indicators impo...