It is widely known that various types of human errors occur in the confused state of a plant operator's mind under abnormal situations. In order to understand the generation mechanism of these human errors in the circumstances of disturbances and equipment failures, we have developed a cognitive information-processing model incorporating the operator's physical and mental states. The operator model is installed on a PC and connected to a boiler plant simulator for training through a human-machine interface model that we also developed. Using this simulation environment we examine perceptual error in the monitoring mode and cognitive error in fault diagnosis during malfunctions. The simulation results coincide well qualitatively with observations of actual plant operations and simulator training. This operator model can be used to analyze the generation mechanism of various types of human errors from the viewpoint of cognitive information processing. Accordingly, we can study how to prevent human errors and evaluate the effects of various measures by running simulations. (C) 2004 Elsevier Ltd. All rights reserved.