In the massive activities of cities, rush hour is temporary while hot roads are congested seriously. Traffic flows are unbalanced severely in two aspects of time and space. Traffic control program is key point while few researches put emphasis upon it. In this paper, a bi-level programming model to make traffic control measure is brought forward. In upper level, providing key member with high transportation service is first thought over, and then the sum of travel delay of the masses and cost of traffic control is minimized. In lower level, evacuation delay of activities personnel is designed to ensure no traffic jam emerged. Rough set genetic algorithm is employed to solving bi-level programming problem in the paper, and an example of traffic control to massive activity verifies the validity of the proposed model.