In the probabilistic Safety Assessment (PSA), probabilistic, statistic, and reliability theory are used to evaluate the risk of nuclear power plant and other large complex production facility. The results of PSA can be used to evaluate the relative importance of subsystems, component, and human actions to the risk of the facility. These importance measures provide a guidance to improve the system design, to allocate the maintenance resources, to plan the training activities. Human reliability analysis plays an important role in the PSA. In this paper, a software based on the Human Cognitive Reliability (HCR) model was developed to predict human error probabilistic. In the software, the performance shaping factor required in the HCR model is determined using fuzzy theory. The data embedded in HCR model is used to generate the human error probability using neural network technique. The export judgements are implemented in the software to simplify the human reliability analysis. This model is validated using human error data from the simulator of a nuclear power plant.