Fisher Linear Discriminant Analysis (LDA) has recently been successfully used as a data discriminantion technique. However, LDA-based face recognition algorithms suffer from a small sample size (S3) problem. It results in the singularity of the within-class scatter matrix S w . To overcome this limitation, this paper has developed a novel subspace approach in determining the optimal projection. This algorithm effectively solves the small sample size problem and eliminates the possibility of losing discriminative information.