Objective Exploring the imaging manifestations and reasons for missed diagnosis of endometrial polyps (EP) in vaginal three-dimensional ultrasound examination. Methods 186 patients who visited our gynecology department from June 2022 to August 2023 due to abnormal uterine bleeding were selected. All patients underwent vaginal three-dimensional ultrasound examination, and the examination results were compared with surgical pathological results to analyze the diagnostic value of vaginal three-dimensional ultrasound examination. Divide into missed diagnosis group (n=32) and no missed diagnosis group (n=85) based on the missed diagnosis situation. Compare the general information and ultrasound characteristics of two groups, use multiple logistic regression analysis to identify the factors for missed diagnosis, construct a predictive model, and evaluate the discrimination and accuracy of the model. Results 85 cases were diagnosed with EP by vaginal three-dimensional ultrasound examination. The ultrasound images of most patients showed clear boundaries between the endometrium and adjacent muscle layers, with papillary echogenicity protruding into the uterine cavity on the inner wall, mostly strong echogenicity, accompanied by punctate or short strip color blood flow signals and moderate resistance arterial blood flow frequency spectrum. 117 out of 186 patients were diagnosed with EP through pathological examination, accounting for 62.90%. 85 cases were diagnosed with EP by vaginal three-dimensional ultrasound, and 32 cases were missed. The total compliance rate of vaginal three-dimensional ultrasound examination is 80.11%, the sensitivity is 72.65%, and the specificity is 92.75%, which is consistent with pathological examination. The ROC curve shows that the AUC of vaginal three-dimensional ultrasound examination is 0.752 (95% CI: 0.726-0.794), which has high diagnostic value. There were statistically significant differences (P<0.05) between the missed diagnosis group and the non missed diagnosis group in terms of intrauterine devices, uterine fibroids, adenomyosis, pregnancy, parity, number of lesions, endometrial thickness, polyp diameter, internal echo of lesions, boundary between endometrium and adjacent muscle layers, and blood flow grading. The intrauterine device, polyp diameter<10 mm, and uneven internal echo of the lesion are independent risk factors for missed diagnosis in vaginal three-dimensional ultrasound examination (P<0.05). The equation for constructing a predictive model for missed diagnosis in vaginal three-dimensional ultrasound examination is P=ea/(l+ea), a=-6.972+1.292 × Is it an intrauterine device (1 or 0)+1.138 × Is polyp diameter<10 mm (1 or 0)+0.969 × Whether the internal echo of the lesion is heterogeneous (1 or 0), the model has good discrimination and accuracy. Conclusion Vaginal three-dimensional ultrasound examination has high clinical application value in the diagnosis of EP. In most patients, ultrasound images show clear boundaries between the endometrium and adjacent muscle layers, with papillary echogenicity protruding into the uterine cavity on the inner wall, mostly strong echogenicity, accompanied by punctate or short strip color blood flow signals and moderate resistance arterial blood flow spectrum. However, for patients with intrauterine devices, polyps with a diameter less than 10 mm, and heterogeneous internal echoes of the lesion, misdiagnosis may occur. For such patients, other auxiliary examinations should be combined for diagnosis.