Abstract: A Kernel Uncorrelated Discriminant Subspace (KUDS) method based on Generalized Singular Value Decomposition (GSVD) for radar target recognition is proposed. The new method combines with the advantage of GSVD and kernel trick, which can effectively overcome the limitation of traditional linear methods in solving singular problem, but also improve the class separability further. In addition, a conclusion from Fishers criterion that there exists no useful discriminative information in the null space of the range profile population scatter matrix is derived, which can be used to reduce the dimensionality of original scatter matrices as well as the computation complexity of the following operation of solving kernel optimal discriminant vectors. Experimental results based on three measured airplanes data confirm the effectiveness of the proposed method.