The implementation of a decentralized model reference adaptive control scheme for a two-axis robotic manipulator is described. It is shown, through an extensive series of experiments, that very good model-following performance and axis decoupling is achieved regardless of manipulator payload and working position.