We document that business cycles of U.S. Census regions are substantially more synchronized than those of European Union countries, both over the past four decades and the past two decades. Data from regions within the four largest European countries confirm the presence of a European border effect – within-country correlations are substantially larger than cross-country correlations. These results continue to hold after controlling for exogenous factors such as distance and size. We consider the role of four factors that have received a lot of attention in the debate about EMU: sectoral specialization, the level of trade, monetary policy and fiscal policy. We find that the lower level of trade between European countries, and to a lesser extent the higher degree of sectoral specialization, can explain most of the observed border effect.