This paper is concerned with the maximum tracking errors in PD-controlled robotic manipulators. In other words, a method for designing a linear tracking proportional-derivative (PD) controller is proposed in this paper, by which the closed-loop system consisting of a robotic manipulator and the PD controller is to be stable in the presence of bounded unknown disturbances and/or model uncertainties. This paper further provides a method of the norm analysis of the associated tracking errors, which corresponds to evaluating the maximum value of the errors in the time-domain. The the validity of the proposed method is demonstrated through a experimental studyx with a typical 3-degrees of freedom (3-DOF) robotic manipulator.