The contents of this paper lay down recommendations for carrying out backtesting procedures in order to determine the accuracy and robustness of bank's internal models for measuring market risk capital requirements. These backtesting procedures typically consist of a periodic comparison of the bank's daily value-at-risk measures with the subsequent daily profit or loss ('trading outcome'). The procedure involves calculating and identifying the number of times over the prior 250 business days that observed daily trading losses exceed the bank's one-day, 99% confidence level VaR estimate (so-called 'exceptions').