A decentralized model-based control method for stabilization and tracking control of a class of large-scale interconnected nonlinear uncertain systems is proposed, following a generalized model-based control approach for robot manipulator control. The subsystem interconnections satisfy the ‘matching conditions’, and may be nonlinear and uncertain. Three controllers are derived, with a built-in trade-off between simplicity of implementation on the one hand and accuracy of tracking/stabilization on the other. In the independent servo control method, the subsystem interactions are treated as disturbances, so possibly large gains are required under purely local feedback. In the offline feedforward compensation method, feedforward compensation is applied for nominal interactions, so much lower local feedback gains are required. In the online feedforward compensation method, the compensation is made online, improving closed loop system performance further. Local bounded stability of the interconnected system under the proposed controllers is shown. Simulation results are presented to illustrate the effectiveness of the proposed method.