According to the theories of financial economic, the standard deviation of asset return is defined as risk measure. This risk measure includes both undesirable and desirable risks (total risk). But, in new financial theories, undesirable risks are more important than desirable risks, because these events ruin companies (and destruct their capital). One method of risk measurement is value at risk (VaR) for measurement of these risks. Of course, there ar three main methods for measuring of risk that includes: a. standard deviatio, b. coefficient of variation and c. VaR. In classic theories of portfolio, the optimal investments weights were determined by minimizing portfolio return variance, but, in new financial theories, minimizing VaR is used for determining these weights. In fact, the principle goal of this paper is introduction of VaR as a suitable method to calculate asset risk and optimal allocation of investment. In this paper, we initially define the VaR method and, consequently, calculate it by using of variance-covariance method based on ARCH and GARCH models. We have assumed that this portfolio includes stocks of 4 companies; therefore, we have calculated optimal weights of investment on these stocks in Tehran Exchange Market. In finally, we have cocluded that, typical investor must allocate highest and lowest wieght for the stock of Bank Melli Investment Co. and Qadir Investment Co., respectively.