Abstract: Objective To determine the validity of self-reported hypertension and its associated factors in Chinese elderly populations. Methods Self-reported data of 13 610 elderly participants of the China Health and Retirement Longitudinal Study were compared with their actual blood pressure measurements. The sensitivity and specificity of self-reported hypertension were calculated corrected for current medication use. Multivariate logistic regression analysis was performed to identify individual characteristics that independently predicted the accuracy of self-reports. Results Overall, self-reported hypertension had low sensitivity (56.3%) and high specificity (96.9%). Agreement between self-reported data and medical records was moderate (Kappa=0.57). The multivariate analysis identified age, sex, education, financial situation, health service utilization, history of diabetes and types of medical insurance as significant factors that determined the validity of self-reported data. Conclusion Self-reporting leads to underestimation of the prevalence of hypertension. The accuracy of self-reporting differs among subpopulations.