More than ten years after the seminal paper by Borjas and Bratsberg (1996) modeling the impact of skills on remigration the empirical evidence on that theory is still mixed. Our paper is to shed light on that issue. Using the GSOEP we test two hypotheses derived from Borjas and Bratsberg (1996) while allowing for endogeneity of host country specific capital. Our results give strong support for their theory. Additionally a sensitivity analysis shows that the insignificance of education in previous studies is due to the test design conducted and cannot be interpreted as falsification of Borjas' and Bratsberg's (1996) theory.