Abstract: This paper proposes a new algorithm,named Kernel Inverse Orthogonalized Fisher Discriminant（KIOFD）,to extract optimal discriminant feature,and applies this method to face recognition.There are two problems in linear face recognition:the first one is that the distribution of face images with different pose,illumination and face expression is complex and nonlinear.The second one is the Small Sample Size（S3） problem.This problem occurs when the number of training samples is smaller than the dimensionality of feature vector,which results in a sigular within-class scatter matrix.For the former,kernel technique can be used to extract nonlinear feature,and for the latter,an inverse fisher discriminant criteria combined with orthogonalized technique is introduced to overcome S3 problem.Three databases,namely ORL,Yale Group B,and UMIST are selected for evalution.The results are encouraging.