This paper proposes a two-step procedure for systemic risk estimation under the stochastic volatility/correlation models. The first step utilizes Fourier transform method for dynamic volatility matrix estimation, and the second step develops efficient importance sampling estimators for extreme event probability. For the empirical analysis, we find that the systemic risk can be useful to measure the stability of financial system because it seems be able to provide early signs for institutions in the U.S. during the 2008- 2010 financial crisis. systemic risk (SRISK) in China and Taiwan are also compared with their financial leverage.