This study tests for the long-run purchasing power parity hypothesis for a panel of 17 African economies, all belonging to the Common Market for Eastern and Southern Africa, over the period 1970–2007. This assumption is considered both in its traditional view, that is, the reversion of the real exchange rate (RER) to a constant mean, and in the Balassa-Samuelson view, that is, the reversion of the RER to a constant trend. For this purpose, we use the new panel data stationary test advocated by Carrion-i-Silvestre et al. that accounts for multiple structural breaks and cross-sectional dependence. Finally, our results give strong evidence for purchasing power parity when the Balassa-Samuelson version is considered. This finding supports the view that purchasing power parity is most likely to be found among high inflation less developed countries and non-CFA countries. Copyright © 2009 John Wiley & Sons, Ltd.