<p>The control of a three-axis parallel mechanism suffers from problems of nonlinearity, strong coupling, and time-varying uncertainties of each motor, which present technical challenges to the dynamic modeling and motion control of the mechanism. This paper proposes a new model-free decoupled and robust repetitive controller to improve the motion trajectory tracking accuracy. Based on the time-delay estimation (TDE) technique, this method can solve the problem of missing parameters of the complete dynamics of the mechanism and realize efficient decoupled control of the mechanism through indirectly applying the overall dynamics to the control structure for closed-loop motion control. To reduce the TDE error, a robust repetitive controller is implemented in the joint space of the mechanism to improve the tracking accuracy of the mechanism with no need of any external sensors. The closed-loop stability of the proposed controller is demonstrated using the Lyapunov approach. Simulations and experimental studies are conducted on the three-axis parallel mechanism. Through comparison with other methods, the results validate that the proposed controller presents better performance in terms of trajectory tracking accuracy and end-effector pose stabilizing ability.</p>