Unknown periodic disturbances are widely found in practical control engineering. A least-mean-square-based adaptive estimator (LMS-AE) is developed to cope with this kind of disturbances. Enlightened by the LMS adaptive filter, an LMS-AE is devised to iteratively learn the amplitude and phase of the components of the periodic disturbances, after the frequency characteristics are extracted by a data-driven technique. Then, the output error between the real system and the observer is introduced to supervise the learning process of the pending weights by the LMS algorithm. Moreover, the convergence condition of the weight updated law is given. The stability conditions of the LMS-AE-based closed-loop control system are analyzed by the separation theorem. The effectiveness and superiority of this proposed control method are verified by a case study and comparisons with other methods.