This paper investigates the long run behavior of real exchange rates in nineteen countries of Latin America over the period 1970 - 2006. Our data does not support the Purchasing Power Parity (PPP) hypothesis, implying that real shocks tend to have permanent effects on Latin America's real exchange rates. By exploiting the advantage of non stationary panel econometrics, we are able to determinate factors that drive real exchanges rate in the long run : the Balassa-Samuelson effect, government spending, the terms of trade, the openness degree, foreign capital flows and the de facto nominal exchange regime. The latter effect has policy implications since we find that a fixed regime tends to appreciate the real exchange rate. This finding shows the non neutrality of exchange rate regime regarding its effects on real exchange rates. We also run estimations for country subgroups (South America versus Caribbean and Central America). Regional results highlight that several real exchange rates determinants are specific to one geographic zone. Finally, we compute equilibrium real exchange rate estimations. Two main results are derived from the investigation of misalignments, [i ] eight real exchange rates are quite close to their equilibrium level in 2006, and [ii ] our model shows that a part of currencies crises that arose in Latin America was preceded by a real exchange rate overvaluation.Facultad de Ciencias Económicas