Abstract: Frequency domain LMS based adaptive beamforming algorithm is presented in this paper. The received signal of array antennas is analyzed, which shows that different DOAs correspond to different frequency, and the received signal is the addition of different spatial frequency signals, Which is the foundation of frequency LMS beamformer. This novel adaptive beamforming algorithm uses FFT as the preprocessing, and uses LMS algorithm to implement adaptive beamforming in frequency domain. Fast fourier transformed signal has a better decorrelation and leads to better convergence of LMS. Theoretical analysis and simulations demonstrate that this algorithm converges faster than the conventional adaptive beamforming algorithm, and it has the better performance. And this algorithm is robust to coherence source. Finally the algorithm is less complex and can be implemented easily, and it can be used widely.