In this work, the problem of trajectory tracking in uncertain underactuated systems is considered. To solve it, a combination of differential flatness and active disturbance rejection control (ADRC) is proposed. The controller design is synthesized in the absence of detailed knowledge of the system model and focuses on dealing with over-amplification of measurement noise, typically seen in conventional single high-gain observer-centered control approaches. The introduced solution is based on fully utilizing the information already available about the governed system, without the necessity for additional measurement devices. To be easily implementable, it is expressed in an industry familiar error-based form with a straightforward tuning method. Through experimental verification, the proposed approach is shown to enhance the disturbance-rejection capabilities of the standard ADRC structure and reduce its sensitivity to measurement noise, thus increasing its practical appeal.