The paper wants to understand the assimilation pattern of foreign migrants in Italy. Three novelties characterize this study. First, the research compares the wage assimilation of international migrants with both internal migrants and local natives in Italy, a country with substantial internal and international migration. This comparison, never exploited before, provides indirect evidence for the role played by language and knowledge of social capital in the assimilation of foreign migrants relative to both natives and internal migrants. Second, we inquired into the possible causes of under-assimilation by controlling for the date of entry and migrant sector concentration. Third, we model new corrections of the selection bias due to return migration. The correction for the selection bias is introduced in the wage equation through a duration extension of the traditional Heckman correction term and alternatively through a hazard rate correction. The empirical test uses the Italian administrative dataset on dependent employment (WHIP), to estimate a fixed effect model for the weekly wages of males aged 18-45 with controls for selection in return migration and unobserved heterogeneity. The three groups of workers start their careers at the same wage level. But, as experience increases, the wage profiles of foreign nationals and natives, both internal migrants and locals, diverges which seems to hint at the importance of language and social capital. However, sector-by-sector analysis shows that in “migrant intense sectors” internal migrants and locals have the same wage profile as foreign workers. Positive selection in returns reinforces the view that the best leave because they have few career options. Thus under assimilation is caused more by community and job segregation than by a lack of language and social capital: alternatively it is the result of their interrelations.