The program evaluation and review technique (PERT) network model is commonly used in project schedule management, in order to consider the influence of uncertainties in production, site conditions, etc. In the PERT network model, an important problem is to estimate the probability of the project completion time over target deadline as a measure of project schedule risk. In the paper, based on the domain decomposition method, a novel simulation method is proposed to estimate the concerned probability. In the method, corresponding to each path in the PERT network, a basic event is defined as the duration of this path over target deadline. Thus the event of the project completion time over target deadline can be expressed as a union of the defined basic events. In the proposed method, a reduction factor is estimated to measure the overlapping between different basic events, and then the concerned probability is estimated by the product of the reduction factor and the sum of the probabilities of basic events. As shown in the illustrative example, the proposed simulation method has good estimation accuracy, and it is more computationally efficient than the Monte Carlo simulation.