In this paper we empirically test three of the most significant theories about the emergence of a European identity. The three approaches considered here are, respectively: first, a "cultural" theory, which understands identities as being based on ethno-cultural factors generated through a long-term (historical) process; second, an "instrumental" theory, which conceives of identities as being based on self-interested calculation (whether economic or political); and a third "civic" theory, which understands identities as being based on agreement over rules for peaceful political co-existence. Our empirical test of these theories exploits Eurobarometer data. In recent years, many researchers have become increasingly dissatisfied with the way these surveys poll attitudes towards the EU. We have contributed to this debate by designing special new questions to measure national and European identities which were included in Eurobarometer 57.2 and are used here for this analysis. Our results provide only partial support for the theories mentioned above. We find that national and European identities are compatible. This is, in part, because while national identities are largely "cultural", European identities are primarily "instrumental". However, we also find that there is a sufficient European common "cultural" ground for a European identity to emerge. We have also confirmed that, because national and European identities are different, the development of a European identity does not necessarily imply the transfer of loyalties from the national to the supranational level. In all the countries analysed here, attachment to the nation remains strong, and certainly greater than attachment to Europe. We also show that it is harder for a European identity to develop in countries with a strong sense of national pride.