Abstract: Delay and its compensation are key problems for Real-time Hybrid Testing (RHT), which was developed from the pseudo-dynamic testing method. Conventional compensation methods are mostly based on constant delay assumption. However, system delay changes during tests due to system nonlinearity and performance change of the specimen, which results in undesirable test results. This paper proposes an adaptive delay compensation method based on online model parameter identification for RHT. The proposed method uses discrete models to simulate the servo loading system, online parameter identification to attain the loading system state; then the system delay is compensated based on this model. This paper firstly describes the method, and then numerically validates the feasibility and the effectiveness. The effect of different models is demonstrated as well.