The thesis investigates the adaptive control of hydraulically-actuated manipulators using a Generalized Predictive Control (GPC) algorithm. The feasibility of applying GPC to a two-link hydraulic manipulator is first studied through computer simulation, and its control performance is compared with that of the well known adaptive Minimum Variance Control (MVC) algorithm. Issues relevant to position and force controls are addressed. Experimental study on a single hydraulic actuator is then carried out on both position and force control. The work consists of the following main parts: (1) A linear mathematical plant model is established suitable for the control equation formulated in single-input single-output (SISO) GPC algorithm. Comprehensive study is conducted to find the effect of design parameters and to test the adaptability of the algorithm through computer simulation. Computer simulation results of minimum variance control are also compared with those belonging to GPC to identify their respective characteristics with the emphasis on adaptability. (2) SISO-GPC algorithm is extended to multiple-input multiple-output (MIMO) GPC algorithm, paying attention to the interaction between links in order to improve the response. Consequently, the control is performed in the Cartesian space instead of the joint space. (3) he adaptive control strategy using SISO-GPC algorithm is then applied to the force control of the manipulator after the establishment of the system model theoretically. Finally, MIMO-GPC algorithm is adopted towards position/force of the manipulator. (4) The efficiency of adaptive control using SISO-GPC algorithm is verified by experimentation, performed on a hydraulic actuator. The results are also compared with those belonging to MVC algorithm. (Abstract shortened by UMI.)