This study aims to expose a possible dynamic structure and nonlinear relationship in exchange rates. Specifically, the analysis derives on the filtered version of USDTRY daily log returns, where TRY stands for the New Turkish Lira with respect to the US Dollar (USD) between August 2001 and February 2007. A carefully applied projective filtering methodology removed most of the noise contaminant. The computation of correlation dimension and Largest Lyapunov Exponent (LLE) supported by the surrogate data testing procedure showed that the nature of the governing dynamics of the filtered series has a significantly different behavior from a stochastic system. All computations support the evidence of deterministic chaos in the reconstructed phase space of the filtered USDTRY log return series. (C) 2009 Elsevier Ltd. All rights reserved.