Prior data on long-term association between legume consumption and hypertension risk are sparse. We aimed to evaluate whether total legume and subtype intakes prospectively associate with hypertension incidence among 8758 participants (≥30 years) from the China Health and Nutrition Survey 2004–2011. Diet was assessed by interviews combining 3-d 24-h food recalls and household food inventory weighing at each survey round. Incident hypertension was identified by self-reports or blood pressure measurements. We applied multivariable Cox regressions to estimate hazard ratios (HR) with corresponding 95 % CI for hypertension across increasing categories of cumulatively averaged legume intakes. For 35 990 person-years (median 6·0 years apiece), we documented 944 hypertension cases. After covariate adjustment, higher total legume intakes were significantly associated with lower hypertension risks, with HR comparing extreme categories being 0·56 (95 % CI 0·43, 0·71; Ptrend < 0·001). Then we found that intakes of dried legumes (HR 0·53 (95 % CI 0·43, 0·65); Ptrend < 0·001) and fresh legumes (HR 0·67 (95 % CI 0·55, 0·81); Ptrend < 0·001) were both related to decreased hypertension hazards. However, further dried legume classification revealed that negative association with hypertension substantially held for soyabean (HR 0·51 (95 % CI 0·41, 0·62); Ptrend < 0·001) but not non-soyabean intakes. In stratified analyses, the association of interest remained similar within strata by sex, BMI, physical activity, smoking and drinking status; rather, significant heterogeneity showed across age strata (Pinteraction = 0·02). Total legume consumption among the over-65s was related to a more markedly reduced hypertension risk (HR 0·47 (95 % CI 0·30, 0·73); Ptrend < 0·001). Our findings suggest an inverse association of all kinds of legume (except non-soyabean) intakes with hypertension risks.