In order to relieve delays at the intersection caused by excessive left-turn traffic，an intersection signal optimization method based on dynamic control of reverse variable lanes was proposed. First，the dynamic control scheme of reverse variable lanes was analyzed based on the realtime data from the detector. Then，with the effective green light time of the phase and the number of variable lanes as constraints，the optimization model with the objective function of the minimizing delay of vehicles at the intersection was established. Next，the intersection signal optimization plan was obtained by substituting real-time data to solve the model. Finally，the case was verified by simulation based on the VISSIM software. The results show that the average delay of vehicles at intersections has been reduced by 14.99%，the number of parking times for left-turn vehicles and straight-going vehicles has been reduced by 21.60% and 38.25% respectively.The average queue length has been reduced by 44.07% and 60.20% respectively，and the maximum queue length has been reduced by 33.85% and 19.63% respectively. It is proved that the optimization method in this paper can significantly improve the operation efficiency of the intersection.