Seismic probabilistic safety assessment (PSA) is developed to give the insight of nuclear power plant risk under earthquake and the main contributors to the risk. However, component failure probability including the initial event frequency is the function of peak ground acceleration (PGA), and all the components especially the different kinds of components at same place will share the common ground shaking, which is one of the important factors to influence the result. In this paper, we propose an analysis method based on Monte Carlo (MC) simulation in which the effect of all components sharing the same PGA level can be expressed by explicit pattern. The Large LOCA accident in AP1000 is analyzed as an example, based on the seismic hazard curve used in this paper, the core damage frequency is almost equal to the initial event frequency, moreover the frequency of each accident sequence is close to and even equal to the initial event frequency, while the main contributors are seismic events since multi components and systems failures will happen simultaneously when a high value of PGA is sampled. The component failure probability is determined by uncertainties in PGA and in component seismic capacity, and the former is the crucial element to influence the result.