Analysis of interest differentials in the European Monetary System has suggested a widespread lack of credibility of the Exchange Rate Mechanism in the sense that in most countries the implicit expected rate of devaluation was significant and not far short of inflation differentials is-aZis the deutschmark. We present and test a model in which prices reflect expectations of exchange rate behaviour. The results provide mixed evidence that price-setting in countries which participated in the ERM was influenced by the exchange rate discipline.* The authors wish to thank the participants of the European Economic Association conference, Prague, September , and two anonymous referees and an associate editor, for comments and suggestions, although responsibility for errors and omissions remains our own.[ ] Kingdom, which functions as a ' control ' since it did not participate in the ERM in this period.Previous work on EMS credibility using inflation data encompasses a variety of approaches to the question (see Blackburn et al., ; Giavazzi and Giovaninni, ). Barrell () looked for structural breaks in wage and price equations in France, Germany, Italy and the United Kingdom, but his results did not yield clear evidence of a regime change at the inception of the ERM. Bleaney () examined the relationship between inflation and central bank independence across OECD countries, and tested whether pegging to the DM by EMS and EFTA countries was equivalent to an increase in central bank independence. He assumed that countries acquired the independence of the Bundesbank in proportion to their refusal to accommodate excess inflation ( is-aZis Germany) through devaluation. His results supported the GP hypothesis. Bini-Smaghi () estimated an inflation regression across OECD countries and found significant negative coefficients for dummies capturing EMS effects. Weber () argued that inflation rates should be stationary only where policy-makers have credibility, and on this basis concluded that only the smaller EMS countries have credibility, and France and Italy do not. This contrasts with the findings of Artis and Nachane (), who concluded that the EMS was credible in France, Italy and the Netherlands, but not Belgium. Kremers () showed that for Ireland, whose currency was pegged to sterling before , inflationary expectations were strongly influenced by expectations of inflation in a basket of ERM countries after , whereas previously they had been largely determined by expected UK inflation. Gressani et al. () also found significant EMS effects for Italy, using the Banca d'Italia model. Artis and Ormerod () tested whether the influence of past German inflation on current domestic inflation rates had increased relative to the pre-ERM period (characterised as -) for France, Belgium and the Netherlands. Unlike most of the other papers their results were negative.Of all the different approaches discussed above, that of Artis and Nachane is closest in spirit to ours....