Although it was widely recognized that the addition of recycled asphalt pavement (RAP) would affect the fatigue performance of hot-mix recycled asphalt mixture (HRAM), a quantitative analysis of how it affected the fatigue performance so far still lacking. The purpose of this study was to identify the effect of RAP content on the fatigue performance of HRAM. For this, the self-developed rejuvenator was used to prepare specimens, then the pavement performance test and the four-point bending test were performed on the specimens to measure the fatigue performance of HRAM. In which, the specimens were prepared with RAP content of 0%, 30%, 40%, and 50% respectively, and the dosage of rejuvenator was 5%. Results suggested that with the addition of RAP, the dynamic stability of HRAM increased by 4.8% ∼ 34.5%, and the rutting deformation decreased by 5.22% ∼ 19.22%. The flexure-tension failure strength of HRAM decreased by 12.39% ∼ 35.66%, and the maximum failure strain decreased by 9.31% ∼ 27.02%. The splitting strength of HRAM decreased by 6.73% ∼ 17.45%, and the TSR decreased by 3.57% ∼ 4.36%. This indicated that the high-temperature performance of HRAM had a positive correlation with the content of RAP, while the low-temperature performance and water stability showed a decreasing trend. The principal component analysis model was established to quantify the influence of RAP content on pavement performance through related performance indexes. From the fatigue performance test, the fatigue life of HRAM was increased with the addition of RAP. The variance analysis under the interaction of multiple factors showed that the p-values for temperature and strain level were 8.14E-4 and 1.418E-20 respectively. The temperature and strain level had a significant effect on the fatigue performance of HRAM, and the strain level was the main factor affecting the fatigue performance, then the fatigue equation and fatigue limit equation based on the temperature factor was established. The effect of different factors on fatigue performance of HRAM was quantified by establishing the back propagation (BP) neural network model, it was found that the prediction value was very close to the test results, and the R2 was 0.98819, which indicated that the established model has good fitting accuracy. The effect of RAP content on pavement failure characteristics was studied by establishing the three-dimensional finite element model, it was found that the deformation depth in each direction had a good linear relationship with the RAP content, and the fitting accuracy was close to 1. A comparison study indicated that the fatigue equation and fatigue limit equation based on temperature factor can better describe the coupling effect between RAP content and fatigue performance, the principal component analysis model can better describe the correlation degree between RAP content and pavement performance, the finite element model can better describe the correlation degree between RAP content and pavement structure deformation.