We introduce pair-copulas model to estimate the Conditional Value at Risk for multivariate stock portfolios.First we model the time series of log-returns of each stock with an ARMA-GARCH model.Then using the pair-copulas model and estimating the parameter of pair-copula.Next,Monte Carlo simulation techniques are applied to obtain the portfolio distribution for the next day, apply our method to different stock portfolios and test the accuracy of the Conditional Value-at-Risk estimates for different confidence levels.The results show that pair-copula constructions, which was based on D-vine, estimated the Conditional VaR at all of the considered quantiles well.