We present the structure and properties of an aggregate model for the European Union, consisting of 34 equations estimated on annual data aggregated over the twelve-member EU, ranging from 1960 to 1997. Estimation was performed by a two-stage procedure taking into account the possible presence of structural breaks in the long-run parameters and allowing for nonlinearity and simultaneity of the dynamic equations. The estimated equations have a very good fit and pass a number of diagnostic tests, suggesting that EU-wide behavioural equations can be defined in a meaningful way. Simulation experiments show that the model is an agile and effective tool for the study of the European economy at an aggregate level. In particular, the model is shown to improve over previous aggregate explanations of the European economy, in that its dynamic multipliers provide a better account of the European unemployment and inflation dynamics in the last decade.