In the most of financial models it’s supposed that distribution of observations is normal and the Value at Risk (VaR) and other criteria of market risk are calculated upon this distribution. This is while observations follow abnormal distributions in reality. So this study calculates Incremental Value at Risk (IVaR) with the assumption of being normal initially and then with regard to real distribution of data and finally compares the results of these two situations. The scope of this study consists of 42 companies present in financial sector of Tehran Stock Exchange during 2009 to 2013.The result show that by using IVaR criterion we can analyze the impact of each stock on creating the risk of portfolio and we can selected the optimal stocks. Also the results confirm this point that analysis of an portfolio’s sensitivity using IVaR criterion and based on that portfolio’s real distribution achieves more accurate and reliable results rather than it’s normal distribution.