In this paper, wavelet packet based FFT and its application to SNR estimation is proposed. OFDM systems demodulate data using FFT. The proposed solution computes the exact result, and its computational complexity is same order of FFT, i.e. O(Nlog2 N). SNR estimation is done inside wavelet packet based FFT block unlike previous SNR estimations techniques which perform SNR estimation after FFT. Wavelet packet analyzed data is used to perform SNR estimation. The proposed estimator is taking into consideration the different noise power levels over the OFDM sub-carriers. The OFDM band is divided into several sub-bands using wavelet packet and noise in each sub-band is considered white. The second-order statistics of the transmitted OFDM preamble are calculated in each sub-band and the power noise is estimated. The proposed estimator is compared with Reddy's estimator for colored noise in terms of mean square error (MSE). © 2008 IEEE.