The assumption of traditional joint normal distribution can not describe tail dependence condition of finance market,which cause low estimation on VaR.This paper uses t Copula to problem and to test 95 percent and 99 percent of VaR of equal weight investment portfolio of the composition of Shanghai stock indicaes and Shengzhen stock indices and the results show that the performance of VaR based on t Copula Model is better than traditional variance-covariance analysis method,historic simulation method,Monte Carlo simulation method and estimation method of correlative coefficient model based on dynamic condition.