We use two variants of Markov switching models to assess changes in output synchronization since the creation of the euro. Out of eight eurozone countries investigated, only one—the Netherlands—has synchronization increased since euro adoption, supporting the ‘endogenous optimal currency area’ argument of Frankel and Rose. However, in three other cases, business cycle synchronization actually fell since the euro's creation. Thus the ‘endogeneity’ of the optimal currency area criteria can go both ways—adopting a common currency may increase synchronization for nations ready for a common currency, but it can lower synchronization for nations that are far from synchronized before monetary unification.