The maximum torque per ampere (MTPA) control is capable of obtaining its maximal ratio of torque to current in a control system of interior permanent magnet synchronous motor (IPMSM). However, when its electrical parameters change with the actual operating conditions, the resulting MTPA trajectory will deflect from the optimal one. To solve this problem, a modified model reference adaptive system (MRAS) method is investigated for the parameter identification of the rotor flux linkage and the stator q-axis inductance, after a tradeoff between the MTPA trajectory derivation degree with parameter change and the rank-deficiency problem in the identification model. In this method, a full-rank estimator and its gain matrix are designed according to the Popov Hyper Stability Theorem. And the current operating point is updated using the identified parameters in order for the real-time tracking of MTPA trajectory. Simulation and experimental results verify that the proposed method enhances remarkably the MTPA tracking control effect and the system’s torque-current characteristics for an IPMSM.