This paper proposes a periodic adaptive disturbance observer (PADOB) to control a PMLSM (Permanent Magnet Linear Synchronous Motor). The PADOB consists of a classical linear DOB and an adaptive mechanism which has been known as the periodic adaptive learning control (PALC). The key idea is to compensate parametric errors between the actual plant and the nominal model of DOB and disturbance forces such as the friction and detent force. The PADOB can improve the instability problem occurred by updating parameters of nominal model in the DOB directly. Through simulation test of the PMLSM, the validity of the PADOB is illustrated.