Based on the data of CHIP 2013 provided by the survey of income of residents in China, this paper adopts Mincer equation in its measurement of educational return rate and its decomposition of wage income gap. It is found that the educational return rate of rural residents is 4.6% after correcting the sample selection bias. Analysis through quantile regression finds that the educational return rate doesn’t increase or decrease linearly with the change of income and the rule of tool variable corrects the endogenous bias estimation. It is also found that the return of academic education presents an “N” shape curve and the congestion of primary and secondary human capital leads to lower return on education of middle income group. Through Shapley regression decomposition, it is found that the difference of human capital stock leads to the main imbalanced distribution of wage income in rural China, while the segmentation of primary and secondary labor markets lies in the economic structure of urban-rural dual segmentation. On this basis, the paper puts forward some countermeasures and suggestions to further reduce the income gap of rural residents as well as to promote poverty alleviation through education.