In practice, it is very significant to monitor the rotor temperature of induction motor quickly and accurately for state evaluation, fault prediction and safety production of electrical equipment. However, due to the limitation of the real operating conditions, it is difficult to directly monitor or indirectly evaluate the rotor temperature. To solve this problem, a rapid parameter identification method for rotor temperature evaluation is proposed. The rotor speed is obtained by using the harmonic identification of rotor slot, and then the rotor resistance is obtained by measuring the electrical parameters at the initial stage of starting which can simulate the locked rotor operation of the motor. The rotor speed and resistance obtained by the above solution are regarded as the extended Kalman filter (EKF) initial value for real-time identification of rotor resistance. The proposed EKF method avoids the problem of long iterative time of initial value, and realizes the rapid on-line identification of rotor resistance. Based on the relationship between the resistance value of metal material and the temperature, the real-time temperature data of rotor can be obtained. Taking a 22 kW induction motor as an example, the rotor temperature under different load conditions is identified, and the validity and feasibility of the proposed method are verified.