Abstract:This paper improves the KMV model, using credit risk premium to measure banks’ credit risk intuitively, and then uses Monte Carlo simulation method to estimate 12 China’s listed banks’ VaR and CVaR. Based on that, it compares the results with those measured by historical simulation method and concludes that: i) historical simulation method overestimates the credit risks faced by the banks; ii) for the sample bank of Bank of China, the extreme credit events are most likely to occur, but for Industrial and Commercial Bank of China, the opposite is true.