This paper proposes a dual-model-based structure for uncertainty attenuation in the trajectory-tracking control of robot manipulators. A conventional model-based control algorithm with an estimated robot dynamic model employed is first placed in the overall structure as a basic control component. To further attenuate the unwanted effects caused by modelling uncertainties, another model-based structure, appended in the outer-loop of the basic control structure as a compensator, is developed based on the concept of the internal model control. Combination of these two model-based structures results in a novel dual-model-based controller for robot manipulators. Sensitivity analyses show that the effects due to both modelling errors and external disturbances can be significantly reduced by applying this dual-model-based control structure without relying on a high-gain control solution. Effectiveness of this control design is successfully demonstrated by simulation studies on a 2 degree-of-freedom robot.