Abstract: In order to improve the control performance under condition of model mismatch, time-varying matrix and observer gain selection, a predictive current control based on an adaptive high-gain observer (AHGOPCC) was proposed in this paper. Firstly, by incorporating parameter mismatches of resistance, inductance and flux, a feed-forward extended model of permanent magnet synchronous motor (PMSM) was established. Secondly, aiming to estimate the current and disturbance state and restrain disturbance of speed variation, a high-gain observer was designed with a standard feed-forward extended PMSM model. Thirdly, an adaptive high-gain matrix was developed to satisfy the performance under full speed operation and improve the observer gain selection. Finally, a discrete predictive current model and a cost function were obtained based on AHGOPCC. The experimental results verify the excellent dynamic performance, robustness and stability of the proposed method.