In order to improve the automatic identification performance of laser print documents, a new sparse representation algorithm based on Gabor features and fisher discrimination kernel dictionary learning was proposed for print documents identification. Proposed method first extracted the image Gabor features, and used kernel principal component analysis to reduce the Gabor features dimension. Based on the fisher discrimination criterion, a dictionary learning method whose dictionary atoms were initialized by the reduced Gabor features was performed. Sparse representation based classification was used to the identification of laser print documents. Experimental results on our database show its efficiency and effectiveness with a correct printer identification rate of 95.79%.