Objectives To assess the clustering of cardiovascular disease (CVD) risk factors in Han and Mongolian adults with prehypertension or hypertension in Northern China. Methods We selected 3227 Han and Mongolian participants (20-80 years old) using a multistage cluster sampling method in 2014. The participants were interviewed by standard questionnaires and underwent anthropometric measurement and biochemical testing. Han and Mongolian participants were divided into optimal, prehypertension, and hypertension groups based on blood pressure. A multinomial logit analysis was performed to explore relationships between CVD risk factor clustering and prehypertension or hypertension, and the heterogeneity between Han and Mongolian was evaluated by the Cochran Q test. The differences between the ethnic groups in the proportions of risk factors was tested with the.2 test. Results The clustering of two or three CVD risk factors in the prehypertension or hypertension groups was consistently higher than in the optimal group (Bonferroni, p<0.0167). The odds ratios (ORs) of prehypertension and hypertension increased with the number of CVD risk factors (p trend <0.0001). In multivariate modelling, the adjusted ORs of one, two, and >= 3 CVD risk factors versus no risk factors was, respectively, 1.95, 2.25, and 2.28 in Han prehypertensive participants, and 1.73, 2.83, and 3.69 in Mongolian prehypertensive participants. In addition, the adjusted ORs were 3.15, 4.75, and 6.49 in Han hypertensive participants, and 1.90, 5.29, and 8.13 in Mongolian hypertensive participants (all p<0.05). There was no significant heterogeneity between Han and Mongolian participants in the prehypertension or hypertension groups. The age-standardised prevalence of >= 3 risk factors was 38.30% in Han men and 39.79% in Mongolian men. The rate was significantly lower in Han women than Mongolian women (9.18% vs 14.55%, p=0.002). Conclusions These findings showed clustering of CVD risk factors in prehypertensive Han and Mongolian adults, and showed prehypertension may be a useful target for intervention.