<p>This study presents a dual-channel disturbance rejection controller for achieving high-precision position control of flexible-joint robots (FJRs) with prescribed performance constraint. The dynamics of the FJR are decomposed into a quasi-steady-state (QSS) model and a boundary-layer model via singular perturbation theory. To effectively address unknown disturbances in the QSS model, a dual-channel disturbance compensator is proposed, which contains a high-frequency disturbance compensator (HFDC) and a generalized proportional integral observer (GPIO). The dual-channel disturbance compensator enables separate estimation of high-frequency and low-frequency components of the lumped disturbance by HFDC and GPIO, respectively. Additionally, the controller for the QSS model is constructed by combining the barrier Lyapunov function and the sliding mode surface, ensuring the exponential stability of the QSS model. The practical exponential stability of the entire system is proven utilizing the extended Tikhonov’s theorem. Numerical simulation and comparative experiments are performed to validate the high-precision tracking performance achieved by the designed approach. To the best of our knowledge, this is the first approach to effectively distinguish and estimate distinct components of the lumped disturbance to achieve precise control of FJRs, particularly in the presence of prescribed performance constraint.</p>