Human model for simulating the cognitive behaviors of Nuclear Power Plant (NPP) operators is constructed and applied for deducing the Human Error Probability (HEP) parameters required for Human Reliability Analysis (HRA) in conjunction with Probabilistic Safety Assessment (PSA) of NPP. Firstly, a framework for describing the basic HEP parameters is proposed. Then, Human Cognitive Reliability (HCR) curves were deduced by analyzing data of a laboratory experiment in which subjects were asked to monitor and diagnose certain abnormal transients in NPP. A human model of NPP operator was constructed for both monitoring and diagnosing the NPP anomaly, based on the experimental observation with applying Reason¡Çs memory model. By comparing two HCR curves, one deduced by human model simulation and the other from experimental data, it was confirmed that the human model simulation could well explain the probabilistic nature of human response in detecting plant anomaly.