The financial market is an environment where changes take place with high velocity. This huge volatility makes necessary the control of the variables involved in formation price process, so losses due to financial transactions can be minimized in a satisfactory way. Value at Risk (VaR) is the most used instrument to do that. VaR summarizes the worst loss, over a target horizon, with a given level of confidence in one single number that reflects the portfolios total variance and the effects of diversification. However, the traditional parametric VaR doesnt consider the liquidity of the portfolio components, and so one can easily underestimate its value at risk. The market considers that financial positions can be sold in any instant of time without impacts to their prices. Nevertheless, in practice, when large amounts are traded there are rigorous impacts in prices. Besides, there is the bid-ask spread commonly observed in every days financial transactions. This dissertation proposes a formula to incorporate liquidity variables when estimating the VaR, using statistic data to do so.