Abstract: A unitary transform based improved Gram-Schmidt orthogonalization (UIGS) adaptive beamforming algorithm is proposed to improve the interference mitigation performance gained from conventional Gram-Schmidt orthogonalization (GSO) adaptive beamformer. Firstly, a unitary transform is performed on the training data vectors and a covariance matrix of the transformed signal is constructed from the transformed data. Then the noise component is subtracted from real-part of the covariance matrix and the resultant matrix is used to reconstruct transformed interference subspace via GSO. Finally the steering vector, on which the identical transformation is performed, is orthogonally projected onto the interference subspace to yield adaptive weighted vector. Computer simulation illuminates that the algorithm has similar performance as the eigenvalue decomposition based subspace projection beamformer in the case of relatively high interference-to-noise ratio (INR).