Traditional signal control systems use prestored timing plans that have been developed offline using historic data. These systems are not responsive to dynamical demand changes of traffic and may deteriorate in performance over time. Dynamic traffic signal control systems can adapt to actual traffic conditions, coping with complex flow patterns and unpredictable variations. They seek continuous optimal system performance. The main goal of this paper is to design a methodology for control of arterial traffic flows. To fulfill this goal, the nonlinear coupled oscillators model is adapted to the traffic signal system of a two-way arterial road. The control methodology is based on measurements of the microscopic occupancy parameters for incoming flows at intersections that have a four-way geometrical structure with four-green splits. The desired signal parameters such as cycle times, green splits, and offsets are adjusted dynamically according to local traffic data. Thus, the desired signal patterns are self-organized through the mutual interactions among the signals. The numerical and case study simulation results demonstrate the effectiveness of the control methodology under the dynamical demand changes of traffic.