||Injection molding is a cyclic process, which consists of three stages: injection, packing-holding and cooling. To ensure the quality of molded parts, it is essential to control the key process variables for each phase. The literature review and industry experience suggest that injection velocity is the key variable to be controlled in the injection phase and nozzle pressure is the key variable to be controlled in the packing-holding phase. Experimental analysis confirms that the injection velocity and the packing pressure are non-linear and time-varying, which makes them difficult to be effectively controlled with a simple traditional controller. This research project shows that a fixed-parameters PI controller fails to effectively control the injection velocity during the injection phase. An adaptive self-tuning regulator (STR) is consequently designed based on a discrete-time dynamic model. It shows better performance than that of the PI controller. Techniques, such as the anti-windup estimation, adaptive feed-forward control and cycle-to-cycle adaptation, are incorporated to enhance the stability and response of the adaptive control system. The controller works well with a wide range of molding conditions. A fuzzy inference system using the nozzle pressure and screw stroke as the inputs is proposed, and designed to detect the proper velocity to pressure control (V/P) switchover. Enhancement is made to ensure the fuzzy system to work effectively with a profiled injection velocity. Extensive experiments have shown that the proposed fuzzy system works well without any adjustment for a wide range of molding conditions, including different injection molding machines. The experimental procedure, similar to that of the injection velocity analysis, is carried out for the nozzle pressure during the packing-holding phase. The adaptive control strategy proposed for the velocity control is also successfully extended to the control of the nozzle packing pressure.