A multi-index comprehensive risk assessment method is proposed to deal with the possible chain failures caused by hidden failures in secondary systems. The Markov state space transfer method is used to solve the time-varying probability and stationary probability of the system in different states. In this paper, a combination of stationary probability and distance protection is used to construct a model to reflect the hidden failures of the system. From three aspects of power grid capacity, power grid users and power grid stability, the risk criteria of isolation, load excision and disassembly of power grid are set up, and the integrated risk criteria is weighted to reflect the cascading failures. A risk assessment system based on Monte Carlo sampling algorithm is established to solve the above criteria, and a specific example is given to demonstrate the multi-criteria risk assessment of the secondary system hidden failures. This work is supported by State Grid Hubei Electric Power Company (No. 26150515003C) and National Natural Science Foundation of China (No. 51207113).