In the past two decades, bearingless permanent magnet synchronous motors (BPMSMs) have been researched widely. Because of the advantages of high efficiency, long operation life, no friction, and so on, the BPMSMs have the prospects to be widely applied in high-speed and high-precision mechanical processing and many other industrial fields. In the control system of the BPMSM, in order to obtain a good suspension performance, the accurate rotor radial displacement detection is necessary. However, the displacement sensors used in the BPMSM system not only increase the cost and the shaft length but also decrease the reliability of the BPMSM. Therefore, the research on rotor radial displacement sensorless control technology is of great significance. In this paper, a rotor radial displacement sensorless control method of the BPMSM based on the model reference adaptive system (MRAS) is proposed. This method not only has higher accuracy and better robustness than the open loop estimation method, but also has simpler structure and lower cost than the high-frequency signal injection method. Firstly, the principle of radial suspension force generation and the mathematical models of the BPMSM are introduced. Secondly, the suspension force windings current model and the actual BPMSM itself are applied as the adjustable model and the reference model, respectively. Then, the adaptive mechanism based on the Popov’s hyperstability theory is derived. The BPMSM rotor radial displacement sensorless control system is designed afterwards. Finally, the comparative simulations and experiments between the proposed method and an open loop estimation method are carried out. The results show that the maximum error between the actual rotor radial displacement and the estimated displacement is 0.005mm, while the data in the open loop estimation method is 0.008mm. In addition, the changes of the parameters in the reference model have little impact on the estimation result.