A range of methods have been developed as project management support. In order to improve decision-making efficiency in conditions of uncertainty, simulations began to be applied in the area of project management. In performing such simulations, the durations of project activities have to be defined as random variables. In this paper it is assumed that activity durations follow a beta PERT distribution. Different information required for decision-making in such a simulation model is obtained by means of adequate statistical methods. The decision-making process can be further improved by modified usage of certain elements of statistical analysis. A computer-aided statistical analysis thus becomes an indispensable tool for a project manager. The construction of adequate decision-making system and the development of software applications are prerequisites for implementing the proposed model.