A data-aided frequency o®set estimation algorithm suited for burst-mode communications is de- scribed and analyzed. The proposed algorithm is based on a Pilot symbol assisted modulation (PSAM) frame struc- ture, which consists of regularly spaced pilot symbols and a group of consecutive pilot symbols embedded in. Firstly, an initial estimate based on consecutive pilot symbols is generated for reducing the phase ambiguity. Secondly, based on both spaced and consecutive pilot symbols, ¯ne preliminary estimates are generated to improve estima- tion accuracy. Lastly, the ¯nal estimate is obtained by combining the ¯ne preliminary estimates with the initial. This three-step algorithm is shown to provide high estima- tion accuracy close to those of FFT-based algorithm at low Signal-to-noise ratio (SNR). In addition, it is suitable for high carrier frequency o®set cases.