Modern research on migration issues occasionally uses linguistic data to make some far-reaching claims on the (lack of) potentials and benefits of multilingualism for the integration of migrants into a host society. Recently, an oft-quoted study by H. Esser assembled and re-analysed some of the most prominent of such studies with the aim to show that language acquisition is only possible at the expense of one of the languages involved, that language acquisition is only effective before the so-called Critical Period, and that multilingualism by and large produces no measurable benefits for society or the multilingual individuals. The present article sets out to re-investigate the studies cited by Esser in terms of their theoretical foundation, validity and methodological scrutiny. It turns out that almost all studies referenced in migration research that make claims about language acquisition and linguistic benefits use vague (subjective) assessments of the learners’ competence in and use of the languages involved. Thus there is little hard evidence to substantiate the sweeping claims made by researchers such as Esser, who base their research mainly on rational choice assumptions. The article therefore argues in favour of developing a productive approach to multilingualism as it is common in current multilingualism, acquisition and sociological research. It draws the attention to both the societal and individual potentials of multilingualism, including the economic benefits to modern societies.