Abstract. We propose steering behavior model to simulate virtual characters (NPC) which walk around in virtual exhibition environment and see exhibits one by one just like humans do. Steering behavior model of such environment has its significance in that it would raise the reality level of virtually recreated exhibition space such as museums or city area with points of attraction. However, currently available steering behavior models and techniques are improper to generate such behavior patterns. Our model is autonomous and individual–centric; virtual characters autonomously determine what exhibit they are going to see, in what order, at which view point, and how long, based on their personal preferences. We also describe the implementation of our model and provide the graphical results of simulations.