The short-term reliability of a power system in the real-time operating state based on EMS/WAMS（energy management system/wide-area measurement system） is studied. The evaluation scheme for online short-term operational reliability including indices， algorithm and application frame is proposed. The key techniques used in the scheme are described. The real-time component models are established to evaluate the system reliability in the current operating state. The time-dependent state probability of a component is adopted to study the short-term reliability of power system. The fast sorting technique is applied to system state selection to balance evaluation accuracy and speed， realizing online evaluation. The selected system states are analyzed based on the system operating conditions during forecasting. The feasibility and validity of the evaluation scheme proposed are proved by application to the IEEE-RTS.