Giant magnetostrictive actuators (GMAs) often work in a close-loop feedback system. This system needs independent sensors which may be difficult to be fixed, besides, excessive sensors may cause more unpredicted problems in a large system. This paper aims to develop a self-sensing GMA. An observer based on piezomagnetic equations is constructed to estimate the stress and strain of the magnetostrictive material. The observer based self-sensing approach depends on the facts that the magnetic field is controllable and that the magnetic induction is measurable. Aiming at the nonlinear hysteresis in magnetization, a hys- teresis compensation observer based on Preisach model is developed. Experiment verified the availability of the observer approach, and the hysteresis compensation observer has higher tracking precision than linear observer for dynamic force sensing. Giant magnetostrictive actuators (GMAs) often work in a close-loop feedback system. This system needs independent sensors which may be difficult to be fixed, besides, excessive sensors may cause more unpredicted problems in a large system. This paper aims to develop a self -sensing GMA. An observer based on piezomagnetic equations is constructed to estimate the stress and strain of the magnetostrictive material. The observer based self-sensing approach depends on the facts that the magnetic field is controllable and that the magnetic induction is measurable. Aiming at the nonlinear hysteresis in magnetization, a hys- teresis compensation observer based on Preisach model is developed. Experiment verified the availability of the observer approach, and the hysteresis compensation observer has higher tracking precision than linear observer for dynamic force sensing.