This article identifies the different ways in which the Eurobarometer surveys measure party attachment in different sets of member states of the European Community or Union. The main variants are an absolute question, an ordinal question and a relative question. Using Eurobarometer data covering the period 1978–94, it is shown that these different questions produce predictably different estimates of the distribution of party attachment. The methodological interpretation is confirmed and a new equivalent measure of party attachment is constructed using the results of a specific test inserted for this purpose in Eurobarometer 41.1. The new measure is constructed in such a way as to preserve the comparability of Eurobarometer data on this topic over time. The new operationalization significantly improves the predictive power of the construct and it is recommended that it be adopted in future.