Abstract: A novel probabilistic safety evaluation model considering human error under cyber attack was investigated in combination with scenarios analysis and emergency operation guidance to study the risk of human error caused by cyber attack on safety digital instrument and control（DCS） system of nuclear power plant. This paper mainly analyzed the operator error behavior caused by cyber attack and provided the corresponding method and process. Then, for a hypothetical accident of loss of secondary loop cold source, a human error risk fault tree model was built to qualitatively analyze the risk of operator error caused by cyber attack and compare the occurrence probability of different operator cognitive competence. Preliminary results of the case study demonstrate that the human error probability cannot be ignored when the operator is obviously misled and the human error risk may be effectively reduced by improving the operator’s cognitive ability to deal the cyber attack events.