This paper explore the long-run absolute purchasing power parity (PPP) hypothesis for a sample of 26 African countries, using both the univariate and the panel unit root tests on annual data for the period 1973-2008. The conventional unit root tests essentially failed to reject the null hypothesis of a unit root in the real exchange rates of the countries which were investigated. Evidence, in favour of PPP for only 7 countries was found. Consequently, a volley of panel unit root tests was employed. The results demonstrated that the null of meanreversion in the real exchange rates of all countries in the sample could not be rejected, implying a breakdown of the PPP in these countries. Therefore, it could be insightful to consider non-linear assessments of adjustment of the exchange rate towards its PPP trajectory. This threshold-type of analysis may convey information useful for policy making.