To overcome the limitations of traditional face recognition methods fo r single sample,a novel method of face recognition based on center-symmetric gradient magnitude and phase patt ern (CSGMP) is proposed. Firstly,gradient magnitude maps and phase maps of a face image are calculated. Secondly,a new operator named center-symmetric local directional pattern (CSLDP) is proposed to encode the g radient magnitude,and gradient phase is quantized into eight regions,then the proposed CSGMP is the combinat ion of the binary codes of phase and CSLDP of magnitude.Finally,CSGMP feature maps are divided into seve ral blocks,and the concatenated histogram calculated over all blocks is utilized as the feature des criptor of face recognition.The recognition is performed by using the nearest neighbor classifier.Experimental results on YALE and AR face databases validate that the CSGMP algorithm is an outstanding method for single sample face recognition under different illumination conditions,different facial expression conditions and par tial occlusion conditions.