This paper examines the long-run determinants of immigration to Germany using a modified version of the Ricardo model. After a brief overview of labour flows to Germany and the related empirical literature, a Ricardian model of migration is estimated using static panel data methods. The results show that variables representing factor abundance appear to have no effect whatsoever on labour flows to Germany in a period of rising unemployment (1974?1999), while variables representing income or productivity differences do have an impact. The latter obtains only if the stock of previous immigrants in the country of destination and the distance between sending countries and the receiving country are controlled for.