In this paper we seek to deepen understanding of out-migration as a social and economic process and to investigate whether cross-sectional earnings assimilation results suffer from selection bias. To model the process of out-migration we conduct a detailed event history analysis of men and women immigrants in Germany. Our 14-year longitudinal study reveals that emigrants are negatively selected with respect to occupational prestige and to stable full time employment. Our results show no selectivity with respect to human capital, earnings, or gender. The likelihood of return migration is strongly determined by the range and nature of social attachments to Germany and origin countries, and grows higher toward retirement. This selective emigration, however, does not appear to distort cross-sectional estimates of earnings assimilation.