We test the assimilation hypothesis as initially proposed by Chiswick (1978) by making. use of a rich panel dataset for Germany which allows us to control for unobserved population heterogeneity and potential selectivity bias arising from an individual's re-migration decision and employment behaviour. To take into account the institutional aspects of the German guest-worker system we use information on an immigrant's expected duration of stay in Germany to distinguish between temporary and permanent migrants and to test for differences in earnings/experience profiles with respect to a foreigner's expected duration of stay. We find that years of schooling in Germany have a strong positive effect on earnings, that earnings/experience profiles of guest-workers differ by expected duration of stay, and that the renumeration of labour market experience in Germany is higher for natives than for most foreigners. The assimilation hypothesis is therefore not supported by the evidence for Germany.