A multiestimation-based adaptive controller is designed for robotic manipulators. The control scheme is composed by a set of estimation algorithms running in parallel. Moreover, each estimation algorithm incorporates a relative dead-zone in order to guarantee adequate estimation properties under the influence of uncertainties in the system's model. Then, a higher order level supervision algorithm decides online the estimator, which parameterize the adaptive controller in real time according to the values of a proposed supervisory index. A minimum residence time between consecutive switchings must be stated in order to guarantee closed-loop stability. It is verified through simulations that the multiestimation-based technique can improve the transient response of adaptive systems by appropriate switching between the various estimation schemes. The robot is controlled by using an impedance-type control. Furthermore, a correcting term based on the difference between the generated torque, which acts as the control signal, and the estimated one, based on the estimation model, has been added to modify the standard control law used in the literature. The closed-loop system is proved to be robustly stable under the influence of uncertainties due to an imprecise modelling of the robotic manipulator.