Experiments on the fatigue crack growth have shown great dispersancy. Study on stochastic crack growth of material at room temperature has been widely performed. However, probabilistic model for crack growth at fatigue-creep has been little investigated due to the complexity of the deterministic model for crack growth at fatigue-creep as well as the time-consuming and the difficulty of the experiments. Traditional crack measurement such as direct current and alternating current electrical potential technique, compliance method is limited for circuit interference at large crack, especially when the temperature is higher than 500°C. Experimental system to achieve real-time FCCG detection at high temperature is established by introducing a long-distance microscope with high magnification and resolution from distances of 15cm to 35cm. The experimental setup consists of a dynamic testing machine, a machine controller, a temperature controlled box, a long-distance microscope and a high temperature furnace from room temperature to 1000°C. Then the fatigue-creep crack growth (FCCG) rate tests on thirty compact tension (CT) specimens made of GH4133B material at 600°C are carried out. The reason for choosing the GH4133B Ni-based superalloy is owing to its popularity in use for the turbine disc of the aero-engine. The tests are conducted on a 100KN capacity servo-hydraulic closed-loop machine employed trapezoidal load with hold time at upon peak load. Based on the crack growth models used for room temperature, the deterministic model for FCCG rate considering the parameters including temperature, hold time is established through comparison of the analytical results with the experimental data. Then the stochastic FCCG model for GH4133B is proposed and the probability of random to reach a specified crack size can be obtained as well as the distribution function of crack size at the service time. Through comparison between the analytical and experimental results, it’s found that the probabilistic FCCG model can fit the experimental data well. Once the stochastic FCCG model is established, it can be used for the probabilistic damage tolerance design of the turbine components made of GH4133B material.