Abstract:High-temperature and high-cycle fatigue tests have been conducted on nickel-based single crystal superalloys at 760℃and 850℃. The macroscopic and microscopic features of the high-cycle fatigue fracture surfaces were assessed by optical and scanning electron microscopy. Fatigue crack propagation tests were conducted to establish the stable propagation characteristics. In addition, a new quantitative analysis parameter for fatigue fracture and a fatigue stress prediction modelling were proposed and applied in the prediction of high-cycle fatigue stress effects. The errors associated with different models are calculated and compared. The results demonstrate that, at 760℃ and 860℃ and a stress ratio (R) of 0.05 and -1, the location of high-cycle fatigue crack initiation, the propagation direction, and the fracture surface characteristics are similar, presenting typical staged characteristics. The cracks all originate from internal porosity and material voids, where the early stage of the crack propagation zone exhibits smooth and flat crystallographic plane characteristics. At the later stage, the crack propagation zone shows "brittle fatigue band" characteristics; the fracture surface is inclined at a certain angle to the loading direction, generating a composite crack. The brittle fatigue band features of the nickel-based single crystal alloys are significantly different from those associated with steel and aluminum polycrystalline alloys in terms of morphology and spacing size order of magnitude. Based on the stable propagation characteristics of the fatigue cracks, the parameters required for the fatigue stress prediction model were calculated. The crack shape factor, "cylindrical semi-elliptical surface crack", which reflects the initiation and propagation direction of the fatigue cracks, was selected. The correction coefficient for the composite crack shape factor that represents the characteristics of the alloy inclined cracks was calculated. Moreover, conventional Paris and Forman models that incorporate the stress ratio and fracture toughness (Kc) were selected as the fatigue stress prediction models. The predicted results were compared with the nominal stress, with a prediction error for the Paris model between 1.0 and 1.9, and between 1.05 and 1.30 in the case of the Forman model. The results indicate that accurate fatigue stress prediction requires the selection of prediction parameters and models which match the fatigue fracture characteristics.