In this paper, the tendency properties of Turkish real effective exchange rate with an inaction band are examined in nonlinear nature. The general SETAR (3) model is used. This is a convenient way of presenting the inactivity band caused by the transfer costs and other factors. In this modelling, the stationarity is defined globally which is allow to unit root in the corridor regime but the outers regimes must be a mean reverting process. The data are used de-meaned and de-trended form. For de-meaned data, we execute both the linearity and the stationarity tests. For de-trended data, only the linearity test is executed. According to our empirical results, the statistical evidence is poor for the validity of PPP in even nonlinear nature. The null of unit root is not rejected for the CPI based reel effective exchange rate and it is rejected only 5% significance level for PPI based one. However, take into account possibility of the existence of trend component, Turkish real effective exchange rates are well characterized by nonlinear process with inactivity band and a tendency property. While these findings support the transaction costs hypothesis, they do not strongly support the validity of the PPP hypothesis.