This thesis will test the accuracy of parametric and non-parametric Value-at-Risk as a risk measure for the swedish stock market. The idea that VaR is a powerful tool of measuring risk that can be understood by all levels of management and regulators has made the approach very popular and is now an industry standard. Many models assume normal distribution and doesn’t take into account the possibility of fat tails. This empirical study of the 30 biggest stocks on the Stockholm stock exchange and the index OMXS30 will show that both parametric and non-parametric VaR underestimates the frequency of VaR breaks.