In this paper, we propose a novel project scheduling method to enhance schedule robustness. The proposed method evaluates and improves a project schedule according to the results of scheduling simulations considering the occurrence of uncertainties. In a series of the scheduling simulation, activities for achieving project’s goal are executed and the existing project schedule is revised by using pre-specified procedure every time an activity delay occurs. It is considered that the simulations including schedule revision tend to be time-consuming for the larger scale project. Our proposed method can drastically reduce a computational time required for evaluation by performing simulations in parallel using GPU. Through the numerical experiments, the computational time for large-scale project is investigated, and effective reduction of a project delay by the proposed method is demonstrated.