This paper proposes an empirical methodology to measure systemic risk by defining it as the contribution of a financial institution to the total capital shortfall of the financial system during a crisis. The authors introduce a systemic risk measure called SRISK, which captures the expected capital shortage of a firm based on its leverage and Marginal Expected Shortfall (MES). MES is the expected loss an equity investor would experience if the market declines substantially. To predict MES, the authors use a dynamic model for market and firm returns, characterized by time-varying volatility and correlation, estimated using TARCH and DCC models. The model allows for flexible inference and accounts for potential tail dependence in shocks. The model is used to forecast MES and SRISK, with the latter being used to rank systemically risky firms. The empirical application on top US financial firms shows that the methodology provides useful rankings of systemically risky firms at various crisis stages. Results highlight the deterioration of financial system capitalization starting in 2007 and indicate that the system was not fully recovered by 2010. The methodology is applied to a sample of 94 top US financial firms between 2000 and 2010, with firms grouped into four categories. The analysis reveals that Broker-Dealers and the Others sector were the most exposed during the crisis. Short-term MES forecasts perform well relative to benchmarks, while long-term MES forecasts provide useful rankings of systemically risky firms. The SRISK index is constructed using long-term MES forecasts, with results showing that eight out of the top ten SRISK firms were troubled institutions one and a half years before the Lehman bankruptcy. The methodology is compared to other systemic risk measures and highlights the importance of time-varying volatility and correlation in MES. The paper concludes that the proposed methodology provides a useful tool for measuring systemic risk and identifying systemically risky firms.This paper proposes an empirical methodology to measure systemic risk by defining it as the contribution of a financial institution to the total capital shortfall of the financial system during a crisis. The authors introduce a systemic risk measure called SRISK, which captures the expected capital shortage of a firm based on its leverage and Marginal Expected Shortfall (MES). MES is the expected loss an equity investor would experience if the market declines substantially. To predict MES, the authors use a dynamic model for market and firm returns, characterized by time-varying volatility and correlation, estimated using TARCH and DCC models. The model allows for flexible inference and accounts for potential tail dependence in shocks. The model is used to forecast MES and SRISK, with the latter being used to rank systemically risky firms. The empirical application on top US financial firms shows that the methodology provides useful rankings of systemically risky firms at various crisis stages. Results highlight the deterioration of financial system capitalization starting in 2007 and indicate that the system was not fully recovered by 2010. The methodology is applied to a sample of 94 top US financial firms between 2000 and 2010, with firms grouped into four categories. The analysis reveals that Broker-Dealers and the Others sector were the most exposed during the crisis. Short-term MES forecasts perform well relative to benchmarks, while long-term MES forecasts provide useful rankings of systemically risky firms. The SRISK index is constructed using long-term MES forecasts, with results showing that eight out of the top ten SRISK firms were troubled institutions one and a half years before the Lehman bankruptcy. The methodology is compared to other systemic risk measures and highlights the importance of time-varying volatility and correlation in MES. The paper concludes that the proposed methodology provides a useful tool for measuring systemic risk and identifying systemically risky firms.