Abstract:OBJECTIVE To explore the application of propensity score matching in cross-sectional study，and to clarify the correlation between drinking and prehypertension. METHODS Based on the baseline survey data of chronic diseases in Xishan Dis⁃ trict，the mixed factor effect between drinking group and no drinking group was balanced by propensity score matching，and the correlation between drinking and prehypertension before and after matching was analyzed. RESULTS The distribution of age，gender， marital status and other factors before matching was unbalanced. After 649 pairs were matched successfully，all the confounding factors were evenly distributed among groups . Before matching，the prevalence of prehypertension in drinking group and no drinking group was statistically significant （χ2=70.212， P＜0.05） . After matching， the prevalence of prehypertension in drinking group was 71.96%，and the prevalence of prehypertension in no drinking group was 68.88%，the difference was not statistically significant （P＞0.05）. After matching，there was no significant difference in prehypertension prevalence between drinking group and no drink⁃ ing group （P＞0.05）. CONCLUSIONS There is no statistical correlation between drinking and prehypertension. Propensity score match⁃ ing can effectively control the confounding bias，and it has a certain application prospect in processing cross-sectional data.