One of the goals of the drivers on the road is to reach destinations in a safe but efficient manner, along with good mood. A driver interacts with other road users in such a way that all of them share the same traffic systems, unavoidably. Unfortunately, most current research has been focused on an individual driver's behavior. Therefore, a main objective of this project is to achieve an understanding of the interaction process between road users and interrelationship between individual behaviors. Meanwhile, despite many advantages that driving simulators can provide better than field study, such as safer, more controllable and repeatable. An important question, however, is whether relying on a single-user driving simulator or standalone systems can give a realistic impression of driving in real life, furthermore whether they can possess high fidelity and validity. Compared with the vehicles operated by real drivers, simplified autonomous vehicles (e.g. pre-programmed and randomly triggered) lack human judgment, intention, adaptability, flexibility, and logic, even though they may have realistic 3D representation in virtual environment. Therefore, in this project, we propose a new driving simulation framework based on a networked multi-users virtual environment. This new simulation prototype allowed the participants to interact with a real driver, and all the subjects’ driving performance data can be collected by the system and transmitted among subjects. With this configuration, multiple subjects’ simulator experiments can be conducted and thus comparable performance data can be obtained to draw meaningful conclusions for interactive driving behavior study.