It is well known that discrete Fourier transform (DFT)-based channel estimators provides good performance with low complexity. Since its performance is influenced by estimated channel taps to lessen the noise effects, it is necessary to estimate the number of the channel taps and the time-domain position of them. Therefore, this paper proposes a modified DFT-based channel estimator with channel impulse response (CIR) adaptation which can simply find the significant channel path delays. The CIR adaptation uses scaling coefficients and threshold level which are generated by Monte-Carlo simulation. Simulation results show that the proposed method has a similar performance of the frequency-domain linear minimum mean squared error (LMMSE) method. Moreover, its performance is rather better than that of the LMMSE method under symbol timing offset (STO).