The present invention relates to an intersection signal control method based on multiple intelligent agents, and belongs to the intelligent traffic intersection control technology field. The method realizes the traffic information real-time dynamic sharing by respectively establishing the intelligent agent models of the traffic entities, such as the vehicles, the road sections, the intersections, the signal lamps, etc., and by utilizing the interactivity and the coordinated cooperation among the multiple intelligent agents, and concretely comprises the three links of the signal cycle optimal regulation and control, the phase conflict green light signal control and the vehicle speed guide auxiliary control. The method can obtain the individual operation state of each vehicle intelligent agent real-timely, and can realize the dynamic interaction between the vehicle intelligent agents and a traffic control system via the speed guide. Meanwhile in allusion to a multi-intelligent agent traffic environment, the caused green light time waste is considered when no vehicle passes at the current phase while a whole phase cycle is optimized, the collision is avoided, and the total number of the vehicles passing the whole intersection is increased by switching to the conflicting phases, the queuing waiting time of the vehicles in phase conflict is reduced, and the average delay of the intersection and the parking times can be reduced effectively.