An intelligent affective tutoring system (IATS) based on back propagation (BP) neural network is developed. Learner's personality data is collected by big-five personality questionnaire. Based on the personality-mood-emotion formula, Emotion data is obtained. Along with the randomness of learner' emotion and complexity of the system, considering the relation of learner' personality, emotion as well as the cognition level, propose an instructional strategies generation model using BP neural network. Training and tests results indicate that the strategies generation model based on BP network has the certain function of associative memory and instructional strategies generated in this way are personalized.