The extraction of facial features points with partial least square regression (Plsr) is a multivariate data analysis technique that can be used to connect between a response variable (Y) to several variables eksplantori (X). Pslr could be efficient in data sets with many variables correlate verry close to and including therein the substantial random noise. This model can extract the relation between facial features and the coordinates of the point of the gray values in the image are used for training. Then this relation is used to determine the location of the facial features in the new test images. the next method is based plsr is single and does not ruquire the other models for different facial characteristic points and all are treared equally. After testing with imagery exercises to get the model, feature extarction can be simplified by using matrix multiplication.