Objective  To investigate and analyze the main clinical risk factors of cystitis glandularis (CG) recurrence after initial treatment, and construct a quantitative prediction model to guide clinical early identification of highrisk patients.Methods  Clinical date of a total of 138 patients with CG confirmed by pathology in our hospital from August 2015 to July 2020 were retrospectively summarized. They were randomly divided into model group (92 cases) and verification group (46 cases ) . Sixty-two patients of recurrence were diagnosed by pathological and clinical symptoms, including 43 patients in model group (46.7%) and 19 patients (41.3%) in validation group. Firstly, the clinical datas between recurrence and non-recurrence patients in model group were compared by single factor, then the main risk factors were screened by multivariate Cox proportional hazard model, and the quantitative prediction model was established according to the weight assignment. The receiver operating characteristic (ROC)curve was used to analyze the diagnostic efficacy of the prediction model in model group and validation group. Finally, the risk of model group and validation group was evaluated according to the prediction model,and KaplanMeier curve was used to compare the recurrence risk during the low,median and high risk patients.Results  Univariate analysis of the model group showed that compared with the patients without recurrence, the patients with recurrence had higher age, longer indwelling catheter time, and higher incidence of urinary tract infection, urinary calculi, squamous metaplasia and atypical hyperplasia (all P<0.05). Cox analysis showed that age, indwelling catheter time, urinary tract infection, urinary calculi, squamous metaplasia and atypical hyperplasia were the main risk factors for CG recurrence (all P<0.05). ROC curve analysis showed that the accuracy of the prediction model in the diagnosis of recurrence in the model group and the validation group was 0.876 and 0.845, respectively (P<0.001). Kaplan-Meier curve analysis showed that the recurrence risk of medium-high risk patients was higher than that of medium-risk patients in the model group and validation group, and the recurrence risk of medium risk patients was higher than that of low-risk patients (all P<0.05).Conclusions  CG has a high recurrence rate of pathological and clinical symptoms. Age, indwelling catheter time, urinary tract infection, urinary calculi, squamous metaplasia and atypical hyperplasia may be the main risk factors to CG recurrence. The establishment of quantitative prediction model has high accuracy and application value for the diagnosis and risk stratification of recurrence.