Blackouts and cascading failures pose a grave threat to the security of power grid, which severely disrupt industrial production and public life. Towards difficulties in blackout prediction, applications of probabilistic model in this area are analyzed. The proposed model consists of admittance model of network topological structure and fault probability model of devices. Admittance model provides a succinct representation of grid architecture, which can be used for discovery of fault devices. On this basis, probability model is used to estimate the blackout probability. Further, processing method are discussed when simultaneous faults and hidden faults are considered. Experimental results demonstrate that the proposed method is of reference value for blackout prediction and the planning of grid.