A widely held nostrum is that countries should exit heavily managed exchange rate regimes when the going is good, rather than when the exchange rate is under pressure to depreciate. Have countries followed this advice in practice? And, if so, how good has the going been? We find that in the past 25 years or so, almost all exits to more flexible regimes were followed by a depreciation of the exchange rate, and that exits were about evenly divided between disorderly and orderly cases. A logit econometric model, indicates that the general circumstances of orderly and disorderly exits have been broadly similar: an overvalued real exchange rate, falling reserves, a difficult fiscal position, and high world interest rates. Wellestablished pegs were less likely to end.