Resumen The aim of this paper is to examine the performance of the Value-at-Risk measure under different distributional models in the highly demanding context of the recent financial crisis. This task is one of the main challenges of the financial industry. In addition to the normal and Student's t distributions, we analyze three distributions especially appropriate for capturing tail risk: The generalized Pareto distribution (GPD), the α-stable distribution, and the g-and-h distribution. We also address the problem of efficiently estimating the parameters of these distributions. Our backtesting analysis shows that GPD and α-stable distributions perform well for this risk measurement purpose. © 2017 Macmillan Publishers Ltd.