Under the natural non-controlled natural environment, one of the difficult problems in the automatic target recognition of aircraft is the attitude variation and occlusion. Therefore, we proposed an aircraft recognition method based on sparse representation. Compared with the existing methods, the highlight of the proposed method is to transform the target image recognition problem into a sparse representation problem of the training samples, which can process the original target images directly, and the processes of feature extraction and selection are not needed. Thus it possesses good robustness to the attitude and occlusion variations of the aircraft targets. Experimental results on 5 kinds of aircraft image data sets show that: the proposed method is feasible, and the recognition rate is above 90%. This method supplies a way for the identification of nonlinear and complex natural air images.