Disaster caused by flood is a very serious problem in Taiwan, and it often bring severe impact by floods, especially for the bridge . We usually use water level as the standard for blocking bridge in the past. In this paper, we issue a new blocking bridge decision for the risk of a bridge, it could provide a more completely method for bridge early warning which combined with high-tech bridge real-time monitoring system. The case for Dung-Shih Bridge is mainly applied in this study. The data for invading typhoon is collected from 2000 to 2009. At first, we use the discharge forecasting system which was developed to estimate Dung-Shih bridge flow for two-steps ahead by using adaptive neuro-fuzzy inference system(ANFIS). Next, disaster risk management is used to be the assessment of the degree of safety bridges by three factors herein. Finally, we make the standard operating procedure(SOP) of real-time monitoring system of Dung-Shih Bridge. This study is expected to be useful for safety warming system of bridges in the future.