Normal distribution of the residuals is the traditional assumption in the classical multivariate time series models. Nevertheless it is not very often consistent with the real data. Copulae allows for an extension of the classical time series models to nonelliptically distributed residuals. In this paper we apply different copulae to the calculation of the static and dynamic Value-at-Risk of portfolio returns and Profit-and-Loss function. In our findings copula based multivariate model provide better results than those based on the normal distribution.