Within the rapidly growing interest in today's robotics industry, modular and reconfigurable robots (MRRs) are among the most auspicious systems to expand the adaptability of robotic applications. They are adaptable to multiple industrial field applications but they also have additional advantages such as versatile hardware, easier maintenance, and transportability. However, such features render the controller design that manages a variety of robot configurations with reliable performance more complex since their system dynamics involve not only nonlinearities and uncertainties but also changing dynamics parameters after the reconfiguration. In this thesis, the motion control problem of MRR manipulators is addressed and hierarchical adaptive control architecture is developed for MRRs. This hierarchical structure allows the adjustment of the nominal parameters of an MRR system for system parameter identification and control design purposes after the robot is reconfigured. This architecture simplifies the design of adaptive control for MRRs which is effective in the presence of dynamic parameter uncertainty, unmodeled dynamics, and disturbance. The proposed architecture provides flexibility in choosing adaptive algorithms applicable to MRRs. The developed architecture consists of high-level and low-level modules. The high-level module handles the dynamic parameters changes and reconstructs the parametric model used for on-line parameter identification after the modules are reassembled. The low-level structure consists of an adaptive algorithm updated by an on-line parameter estimation to handle the dynamic parameter uncertainties. Furthermore, a robust adaptive term is added into this low-level controller to compensate for the unmodeled dynamics and disturbances. The proposed adaptive control algorithms guarantee uniformly ultimate boundedness (UUB) of the MRR trajectories in terms of robust stability despite the dynamic parameter uncertainty, unmodeled dynamics, changes in the system dynamics, and disturbance.