With the arrival of the new era,the behavior of migrant workers returning home is deeply affected by the alienation of their human capital. Based on the classical Todaro model,this paper introduces the relevant variables of human capital and constructed an extended Todaro migration decision model. Through the questionnaire of 526 migrant workers in Shanxi Province,the paper uses the Multinomial Logistic model to analyze human capital’s influence mechanism on labor behavior decision,and further studies the difference of returning home decision between the new generation,the middle generation and the old generation by the sub-sample model in view of the rural migrant workers in three employment flow states,“not going out”,“working without returning” and “migrant workers returning home”. The results of the empirical study verifies that human capital factors have a significant impact on the returning behavior of migrant workers,which is manifested in the fact that migrant workers with low education level,long working years and non-agricultural skills are more likely to return to their hometowns. The sub-sample regression model shows that this returning characteristics of the older generation of migrant workers are more prominent. In addition,the study also finds that women are significantly more likely to make the decision not to go out and to return home,and older migrant workers are more likely to choose to return home. In view of this,the paper puts forward the following countermeasures and suggestions:first,we should increase investment of educational resources in rural areas to raise the educational level of migrant workers; second,we should strengthen vocational training for migrant workers to improve their employability; third,we should consolidate infrastructure construction in rural areasto optimize the environment for migrant workers to return home and find jobs.