The urban traffic flow has complex nonlinear dynamic behavior. It is very difficult to precisely model it in urban traffic control. Meanwhile, the macroscopic traffic flow appears apparent cyclical characteristics in one day cycle. Therefore, an iterative learning control strategy for signal timing of urban regional traffic is proposed. Through iterative control of the traffic signals, the average road occupancy rates in the regional traffic network achieve the desired ones. Thus, the green time is fully utilized and the traffic congestion is effectively prevented, which makes the traffic flow run more efficiently and smoothly in the network. With rigorous analysis, the proposed control scheme guarantees the asymptotic convergence along the iteration axis. The simulation results show the effectiveness of the proposed method.