Central blood pressure level is not always consistent with peripheral blood pressure level, and especially their joint effect on incident hypertension is not well established. A total of 1607 non-hypertensive subjects from an atherosclerosis cohort in Beijing, China were included. Central systolic blood pressure (cSBP) was obtained using Omron HEM-9000AI machine and peripheral systolic blood pressure (pSBP) was measured using Omron HEM-7117 electronic sphygmomanometer, separately. Hypertension was defined as BP ≥ 140/90 mmHg or self-reported hypertension or taking any antihypertension drugs at the follow-up survey. After a median follow-up of 2.3 years, incident hypertension was 13.1%. Every 1 standard deviation increase of cSBP and pSBP was associated with 1.98 (95%CI: 1.69-2.33) and 2.84 (95%CI: 2.30-3.52) times of incident hypertension after adjustment for confounders. Moreover, hypertension risk in single pSBP ≥ 120 mmHg group, single cSBP ≥ 120 mmHg group, and both pSBP and cSBP ≥ 120 mmHg group was 2.83 (95%CI: 0.98-8.16), 3.28 (95%CI: 1.24-8.70), and 11.47 (95%CI: 4.97-26.46) times higher than both pSBP and cSBP < 120 mmHg group, respectively. The joint effect of cSBP and pSBP is superior to either cSBP or pSBP to predict incident hypertension in a Chinese community-based population. Screening of central blood pressure should be considered in non-hypertensive population for the purpose of primary intervention, especially for subjects with pSBP ≥ 120 mmHg.