This paper uses the China Household Income Project 2018 dataset to estimate returns to education for various Hukou-migration subgroups. We overcome the endogeneity problem of years of schooling using an instrument based on the Great Expansion of Higher Education policy. Our results indicate that the highest returns are for urban native workers (27.4%), followed by urban Hukou-converted (25.0%) and rural native workers (14.7%). In contrast, the returns to education for rural-urban migrant workers are insignificant. Further analyses suggest that Hukou conversion significantly increased the returns to education for rural-origin people by enabling them access to better job opportunities.