Macroscopic fundamental diagram (MFD) describes a unimodal, low-scatter relationship between network vehicle density and network space mean flow for homogenous network regions, and provides a possible solution to traffic signal control for urban road networks. Transfer between multiple states has been ignored in a few control methods, despite continuous advances in the areas of congestion perimeter control because different states should have different control objectives. This paper proposes a boundary hybrid control for two-region urban cities based on three network states: free flow, critical flow, and congestion flow as characterized by MFD. The hybrid perimeter control for two-region urban cities with different states can be set up according to different control objectives by analyzing the control adaptation characteristics for the single-control method of Bang–Bang, PI feedback control, and nonlinear PI control based on the MFD curve at a single-boundary entrance under different conditions. Furthermore, the confirmation of rapid reduction in the saturation, traffic distribution, and signal timing of the boundary intersection is derived. An urban center network located approximately 20 km² in Hefei is taken as an example. Comparison results show that the hybrid control performs significantly better than the single-control scheme in completing vehicle travel, decreasing travel delay, and preventing network oversaturation.