The influx of Polish workers into the UK requires a renewal of migration theories and presents important empirical puzzles. Both problems stem from the specificity of this migration wave, which does not correspond to classical models. The inadequacy of classical migration theories is demonstrated by the inaccuracy of forecasts from before 2004. The ‘optimistic’ forecasts (e.g. Boeri and Brücker 2001) foresaw an influx many times smaller than the actual one. But the ‘pessimistic’ forecasts were wrong when expecting a movement towards social benefits (so-called ‘social raids’). The paper shows how the influx of Polish workers is different from previous migration waves: it is, at least in the intention, short-term, and includes a high share of young women. In addition, these workers maintain strong contacts and networks with their home country (frequent travel, new communication technologies), often compare living conditions from different European countries, and show an unforeseen willingness to join local trade unions. The issue is therefore if these people qualify for the definition of (classic) ‘migrants’, or of ‘transnational migrants’ (Pries 2003) or even of ‘cosmopolitans’ (Cohen 2004)? In this way they could, for instance, combine Polish, British, and generally European values on work and social customs. The paper combines the Worker Registration Scheme data with early findings from research (interviews, case studies) on Poles in the West Midlands. This region has the largest number of new Polish migrants in the UK (17% of the total), together with numerous ‘old’ Polish communities. This case is particularly interesting because in this region the share of Poles working through temporary agency employment is the highest (57%). Experiences of both mobilisation and segregation confirm the ambivalent and dynamic nature of this migration wave.