This paper presents an improved command-filtered backstepping control method for a class of multi-source-disturbed nonlinear systems based on an active-uncertainty-estimation-and-compensation approach. Exploiting the available model information of a plant, a reduced-order extended-state observer is designed to estimate a matched total uncertainty. Command filters are constructed to estimate virtual control inputs. Both the estimated states and estimation errors are used for the construction of the backstepping control law, which guarantees the stability of the system and compensates for uncertainties. Moreover, a particle-swarm-optimization algorithm is adopted to simultaneously optimize the state observer gain and backstepping-control gains. Finally, a case study on a rotational system shows that the presented method achieves better uncertainty-suppression and tracking performance in both transient and steady state than other related methods.