Traffic congestion is a substantial problem plaguing modern society. Researchers and practitioners working on transportation systems turn to methods from automatic control to decrease congestion and improve urban mobility. Considering the sheer physical size, complexity of the dynamics, and the interaction between human and machine decision making, modeling and control of traffic in large-scale urban networks remains a challenging problem. Presence of constraints, nonlinear dynamics, and possibility of access to some future knowledge point to the suitability of using model predictive control, which is an advanced control technique based on real-time repeated optimization. Motivated by the potential of integrating the perimeter control type actuation via large-scale route guidance, in this paper the authors develop a nonlinear model predictive control scheme with authority over both types of actuators, for improving mobility in large-scale urban networks. Moreover, a sophisticated large-scale urban network model is proposed, which has the feature of prohibiting cyclic traffic flow, leading to more realistic simulation of urban traffic. Simulation studies with a congested scenario and driver compliance analysis on a 7-region urban network are provided, which indicate that the proposed control scheme, when compared with a perimeter control-only scheme, can achieve substantial improvement in urban mobility, even for modest levels of compliance.