The extraction of effective features is extremely important for understanding the intrinsic structure hidden in high-dimensional data. In recent years, sparse representation models have been widely used in feature extraction. A supervised learning method, called sparsity preserving discriminative learning (SPDL), is proposed. SPDL, which attempts to preserve the sparse representation structure of the data and simultaneously maximize the between-class separability, can be regarded as a combiner of manifold learning and sparse representation. More specifically, SPDL first creates a concatenated dictionary by class-wise principal component analysis decompositions and learns the sparse representation structure of each sample under the constructed dictionary using the least squares method. Second, a local between-class separability function is defined to characterize the scatter of the samples in the different submanifolds. Then, SPDL integrates the learned sparse representation information with the local between-class relationship to construct a discriminant function. Finally, the proposed method is transformed into a generalized eigenvalue problem. Extensive experimental results on several popular face databases demonstrate the effectiveness of the proposed approach.