Dependency between the financial series is a key parameter used to estimate risk models. Value at Risk (VaR) is one of the most important measures used for financial risk management; currently there are different methods to estimate, as the historical simulation method, which assumes no distribution on returns risk factors or assets, or parametric methods that assumes normality on distributions. In this paper we propose the theory of copula as a measure of dependence between sets, we use a ARMA-GARCH-Copula model for estimating VaR, for the specific case we take a portfolio composed of two assets, daily US dollar-peso and Euro-Peso exchange rates.The results show that the estimate of VaR through copulas is more accurate compared to traditional methods.