The invention relates to a preparation and a recurrence risk model for judging the prognosis of early endometrial cancer. The present invention judges the POLE mutation status by sequencing, detects the mismatch repair protein, p53 and VISTA in 4 by immunohistochemical method, and judges the histological grade and immunohistochemical result of the tumor by observing under a microscope, so as to determine the molecular classification of endometrial cancer. Type, tumor grade, and expression of VISTA, and then assign a risk score, 0 for POLE mutation, 1 for dMMR and NSMP, and 2 for p53 in molecular typing; 0 for VISTA-positive and 1 for VISTA-negative 1 point for high-grade tumors and 0 for low-grade tumors. The overall risk score is the sum of the three above-mentioned scores, and the total score ranges from 0 to 4: 0–1 as low risk, 2 as intermediate risk, and 3–4 as high risk. The model can effectively predict the recurrence and death risk of early stage (FIGO I and II) endometrial cancer, so as to avoid over-treatment and under-treatment, and achieve the purpose of precise treatment and individualized treatment.