As a more recently developed dimension of social capital, linking social capital seems to be less frequently discussed in the field of public health. Following the classic definition of linking social capital proposed by Szreter and Woolcock (Int J Epidemiol 33(4):650–667, 2004), this study defined linking social capital from the perspective of political capital. Political trust, political efficacy, political participation and membership in the Chinese Communist Party were used to measure linking social capital. Based on the data from Chinese General Social Survey of 2010 (N = 3209), this study used extended regression model to address potential endogenous problems of linking social capital and estimate its disparities in health return by the level of family income. Results show that only certain associations between linking social capital and psychological health were observed without endogenous problems addressed, and no significant family income disparity in these associations was observed. However, with endogenous problems addressed, it was observed that linking social capital was more beneficial to the health promotion, including both physical and psychological health, of the poor. These results are intrinsically in line with the classic definition of linking social capital, and they also indirectly support the applicability of this classic definition in unique Chinese settings.