This paper examines the factors that determine the long-term real exchange rate.  Is the real exchange rate a stationary variable (fixed and lacking a time trend), as uniform price theory assumes, or does it depend on long-term factors, particularly on relative productivity (Samuelson (1964) and Balassa (1964))?  We examined the change in the real exchange rate during a long period of three decades in a broad panel of countries, and found that the correlation between it and between the change in per capita GDP (which is a proximate estimation of the change in relative productivity) is derived solely from the poor countries.  A further examination, relating to a smaller number of countries and a shorter time frame, did not find a correlation between the development of the exchange rate and the development of the two direct indices of relative productivity: 1) relative manufacturing productivity in relation to overall productivity in the economy; and 2) the technological sophistication of goods exports.  Integration-line tests found that the exchange rate is a stationary variable, and that shocks to the exchange rate are less persistent than shocks to per capita GDP.  An estimation of panel data using the Arellano Bond method, which designed to deal with the high serial correlation of the exchange rate and the problem of endogeneity between it and per capita GDP, finds that the correlation between the two variables in the panel data reflects the short-term development (correlation over the course of the business cycle) and that there is no correlation between the two over the long term.