Abstract: Objective To validate and compare the effectiveness and reliability of different cardiovascular risk prediction models in a follow-up cohort of hypertensive patients. Methods A total of 9 924 hypertensive patients aged 40-70 years who participated in the baseline survey of follow-up cohort study of hypertension in Jiangsu province from January 2010 to December 2010 and followed up to May 2020 were included in this study. China prediction model of atherosclerotic cardiovascular disease risk（China-PAR）, ischemic cerebrovascular disease of 10-year risk assessment in the national population（ICVD） and Framingham Risk Score（FRS） were used to calculate the predicted 10-year cardiovascular risk. Kaplan-Meier method was used to adjust the actual incidence of new cardiovascular events over 10 years, and the ratio of predicted risk to actual incidence was calculated to evaluate the effect of the three models. The prediction reliability of the model was evaluated by Harrell’s C statistics, chi-square value of calibration degree and calibration graph. Results During the mean follow-up of（8.30±1.45） years, a total of 1 017 new cardiovascular events occurred. In male hypertensive patients, the C-statistic and 95% confidence interval（CI） of China-PAR were 0.707（0.693-0.721）, which were better than ICVD 0.656（0.641-0.670） and FRS 0.681（0.667-0.695）, DeLong test, both P＜0.001, and the prediction accuracy was high. Although the C-statistic of China-PAR was greater than ICVD and FRS in women, the discrimination and calibration of the three models were poorer. Conclusions China-PAR has better differentiation than ICVD and FRS in predicting 10-year cardiovascular disease risk in Jiangsu Province, especially in the male population. However, the reliability of prediction needs to be improved and a more appropriate risk prediction model should be established according to the characteristics of hypertensive population, especially those in high-risk groups.